Fix: Move Alpine.js router store registration inline

The spa-router.js was loaded with defer but registered the Alpine.js store using the 'alpine:init' event. Since Alpine.js also loads with defer, there was a race condition where Alpine could initialize before the event listener was registered, causing $store.router to be undefined. Moved the entire router store definition and registration inline in spa.html so it's guaranteed to be registered before Alpine.js initializes. Co-authored-by: mudler <2420543+mudler@users.noreply.github.com>
Fix: Remove talk.js from SPA to prevent null element errors
2026-02-04 11:42:57 -05:00 · 2026-01-11 08:34:47 +00:00 · 2026-01-11 08:10:21 +00:00 · 2026-01-11 07:52:04 +00:00 · 2026-01-11 07:38:52 +00:00 · 2026-01-10 23:54:42 +00:00
145 changed files with 4011 additions and 13702 deletions
--- a/.github/workflows/backend.yml
+++ b/.github/workflows/backend.yml
@@ -105,32 +105,6 @@ jobs:
            dockerfile: "./backend/Dockerfile.python"
            context: "./"
            ubuntu-version: '2404'
-          - build-type: 'cublas'
-            cuda-major-version: "12"
-            cuda-minor-version: "9"
-            platforms: 'linux/amd64'
-            tag-latest: 'auto'
-            tag-suffix: '-gpu-nvidia-cuda-12-qwen-tts'
-            runs-on: 'ubuntu-latest'
-            base-image: "ubuntu:24.04"
-            skip-drivers: 'false'
-            backend: "qwen-tts"
-            dockerfile: "./backend/Dockerfile.python"
-            context: "./"
-            ubuntu-version: '2404'
-          - build-type: 'cublas'
-            cuda-major-version: "12"
-            cuda-minor-version: "9"
-            platforms: 'linux/amd64'
-            tag-latest: 'auto'
-            tag-suffix: '-gpu-nvidia-cuda-12-pocket-tts'
-            runs-on: 'ubuntu-latest'
-            base-image: "ubuntu:24.04"
-            skip-drivers: 'false'
-            backend: "pocket-tts"
-            dockerfile: "./backend/Dockerfile.python"
-            context: "./"
-            ubuntu-version: '2404'
          - build-type: 'cublas'
            cuda-major-version: "12"
            cuda-minor-version: "0"
@@ -150,7 +124,7 @@ jobs:
            platforms: 'linux/amd64'
            tag-latest: 'auto'
            tag-suffix: '-gpu-nvidia-cuda-12-llama-cpp'
-            runs-on: 'bigger-runner'
+            runs-on: 'ubuntu-latest'
            base-image: "ubuntu:24.04"
            skip-drivers: 'false'
            backend: "llama-cpp"
@@ -366,32 +340,6 @@ jobs:
            dockerfile: "./backend/Dockerfile.python"
            context: "./"
            ubuntu-version: '2404'
-          - build-type: 'cublas'
-            cuda-major-version: "13"
-            cuda-minor-version: "0"
-            platforms: 'linux/amd64'
-            tag-latest: 'auto'
-            tag-suffix: '-gpu-nvidia-cuda-13-qwen-tts'
-            runs-on: 'ubuntu-latest'
-            base-image: "ubuntu:24.04"
-            skip-drivers: 'false'
-            backend: "qwen-tts"
-            dockerfile: "./backend/Dockerfile.python"
-            context: "./"
-            ubuntu-version: '2404'
-          - build-type: 'cublas'
-            cuda-major-version: "13"
-            cuda-minor-version: "0"
-            platforms: 'linux/amd64'
-            tag-latest: 'auto'
-            tag-suffix: '-gpu-nvidia-cuda-13-pocket-tts'
-            runs-on: 'ubuntu-latest'
-            base-image: "ubuntu:24.04"
-            skip-drivers: 'false'
-            backend: "pocket-tts"
-            dockerfile: "./backend/Dockerfile.python"
-            context: "./"
-            ubuntu-version: '2404'
          - build-type: 'cublas'
            cuda-major-version: "13"
            cuda-minor-version: "0"
@@ -457,32 +405,6 @@ jobs:
            backend: "vibevoice"
            dockerfile: "./backend/Dockerfile.python"
            context: "./"
-          - build-type: 'l4t'
-            cuda-major-version: "13"
-            cuda-minor-version: "0"
-            platforms: 'linux/arm64'
-            tag-latest: 'auto'
-            tag-suffix: '-nvidia-l4t-cuda-13-arm64-qwen-tts'
-            runs-on: 'ubuntu-24.04-arm'
-            base-image: "ubuntu:24.04"
-            skip-drivers: 'false'
-            ubuntu-version: '2404'
-            backend: "qwen-tts"
-            dockerfile: "./backend/Dockerfile.python"
-            context: "./"
-          - build-type: 'l4t'
-            cuda-major-version: "13"
-            cuda-minor-version: "0"
-            platforms: 'linux/arm64'
-            tag-latest: 'auto'
-            tag-suffix: '-nvidia-l4t-cuda-13-arm64-pocket-tts'
-            runs-on: 'ubuntu-24.04-arm'
-            base-image: "ubuntu:24.04"
-            skip-drivers: 'false'
-            ubuntu-version: '2404'
-            backend: "pocket-tts"
-            dockerfile: "./backend/Dockerfile.python"
-            context: "./"
          - build-type: 'l4t'
            cuda-major-version: "13"
            cuda-minor-version: "0"
@@ -719,39 +641,13 @@ jobs:
            dockerfile: "./backend/Dockerfile.python"
            context: "./"
            ubuntu-version: '2404'
-          - build-type: 'hipblas'
-            cuda-major-version: ""
-            cuda-minor-version: ""
-            platforms: 'linux/amd64'
-            tag-latest: 'auto'
-            tag-suffix: '-gpu-rocm-hipblas-qwen-tts'
-            runs-on: 'arc-runner-set'
-            base-image: "rocm/dev-ubuntu-24.04:6.4.4"
-            skip-drivers: 'false'
-            backend: "qwen-tts"
-            dockerfile: "./backend/Dockerfile.python"
-            context: "./"
-            ubuntu-version: '2404'
-          - build-type: 'hipblas'
-            cuda-major-version: ""
-            cuda-minor-version: ""
-            platforms: 'linux/amd64'
-            tag-latest: 'auto'
-            tag-suffix: '-gpu-rocm-hipblas-pocket-tts'
-            runs-on: 'arc-runner-set'
-            base-image: "rocm/dev-ubuntu-24.04:6.4.4"
-            skip-drivers: 'false'
-            backend: "pocket-tts"
-            dockerfile: "./backend/Dockerfile.python"
-            context: "./"
-            ubuntu-version: '2404'
          - build-type: 'hipblas'
            cuda-major-version: ""
            cuda-minor-version: ""
            platforms: 'linux/amd64'
            tag-latest: 'auto'
            tag-suffix: '-gpu-rocm-hipblas-faster-whisper'
-            runs-on: 'bigger-runner'
+            runs-on: 'ubuntu-latest'
            base-image: "rocm/dev-ubuntu-24.04:6.4.4"
            skip-drivers: 'false'
            backend: "faster-whisper"
@@ -764,7 +660,7 @@ jobs:
            platforms: 'linux/amd64'
            tag-latest: 'auto'
            tag-suffix: '-gpu-rocm-hipblas-coqui'
-            runs-on: 'bigger-runner'
+            runs-on: 'ubuntu-latest'
            base-image: "rocm/dev-ubuntu-24.04:6.4.4"
            skip-drivers: 'false'
            backend: "coqui"
@@ -876,32 +772,6 @@ jobs:
            dockerfile: "./backend/Dockerfile.python"
            context: "./"
            ubuntu-version: '2204'
-          - build-type: 'l4t'
-            cuda-major-version: "12"
-            cuda-minor-version: "0"
-            platforms: 'linux/arm64'
-            tag-latest: 'auto'
-            tag-suffix: '-nvidia-l4t-qwen-tts'
-            runs-on: 'ubuntu-24.04-arm'
-            base-image: "nvcr.io/nvidia/l4t-jetpack:r36.4.0"
-            skip-drivers: 'true'
-            backend: "qwen-tts"
-            dockerfile: "./backend/Dockerfile.python"
-            context: "./"
-            ubuntu-version: '2204'
-          - build-type: 'l4t'
-            cuda-major-version: "12"
-            cuda-minor-version: "0"
-            platforms: 'linux/arm64'
-            tag-latest: 'auto'
-            tag-suffix: '-nvidia-l4t-pocket-tts'
-            runs-on: 'ubuntu-24.04-arm'
-            base-image: "nvcr.io/nvidia/l4t-jetpack:r36.4.0"
-            skip-drivers: 'true'
-            backend: "pocket-tts"
-            dockerfile: "./backend/Dockerfile.python"
-            context: "./"
-            ubuntu-version: '2204'
          - build-type: 'l4t'
            cuda-major-version: "12"
            cuda-minor-version: "0"
@@ -955,32 +825,6 @@ jobs:
            dockerfile: "./backend/Dockerfile.python"
            context: "./"
            ubuntu-version: '2404'
-          - build-type: 'intel'
-            cuda-major-version: ""
-            cuda-minor-version: ""
-            platforms: 'linux/amd64'
-            tag-latest: 'auto'
-            tag-suffix: '-gpu-intel-qwen-tts'
-            runs-on: 'arc-runner-set'
-            base-image: "intel/oneapi-basekit:2025.3.0-0-devel-ubuntu24.04"
-            skip-drivers: 'false'
-            backend: "qwen-tts"
-            dockerfile: "./backend/Dockerfile.python"
-            context: "./"
-            ubuntu-version: '2404'
-          - build-type: 'intel'
-            cuda-major-version: ""
-            cuda-minor-version: ""
-            platforms: 'linux/amd64'
-            tag-latest: 'auto'
-            tag-suffix: '-gpu-intel-pocket-tts'
-            runs-on: 'arc-runner-set'
-            base-image: "intel/oneapi-basekit:2025.3.0-0-devel-ubuntu24.04"
-            skip-drivers: 'false'
-            backend: "pocket-tts"
-            dockerfile: "./backend/Dockerfile.python"
-            context: "./"
-            ubuntu-version: '2404'
          - build-type: 'intel'
            cuda-major-version: ""
            cuda-minor-version: ""
@@ -1041,7 +885,7 @@ jobs:
            platforms: 'linux/amd64,linux/arm64'
            tag-latest: 'auto'
            tag-suffix: '-cpu-llama-cpp'
-            runs-on: 'bigger-runner'
+            runs-on: 'ubuntu-latest'
            base-image: "ubuntu:24.04"
            skip-drivers: 'false'
            backend: "llama-cpp"
@@ -1067,7 +911,7 @@ jobs:
            platforms: 'linux/amd64,linux/arm64'
            tag-latest: 'auto'
            tag-suffix: '-gpu-vulkan-llama-cpp'
-            runs-on: 'bigger-runner'
+            runs-on: 'ubuntu-latest'
            base-image: "ubuntu:24.04"
            skip-drivers: 'false'
            backend: "llama-cpp"
@@ -1408,6 +1252,19 @@ jobs:
            dockerfile: "./backend/Dockerfile.python"
            context: "./"
            ubuntu-version: '2404'
+          - build-type: 'l4t'
+            cuda-major-version: "12"
+            cuda-minor-version: "0"
+            platforms: 'linux/arm64'
+            skip-drivers: 'true'
+            tag-latest: 'auto'
+            tag-suffix: '-nvidia-l4t-arm64-neutts'
+            base-image: "nvcr.io/nvidia/l4t-jetpack:r36.4.0"
+            runs-on: 'ubuntu-24.04-arm'
+            backend: "neutts"
+            dockerfile: "./backend/Dockerfile.python"
+            context: "./"
+            ubuntu-version: '2204'
          - build-type: ''
            cuda-major-version: ""
            cuda-minor-version: ""
@@ -1421,32 +1278,6 @@ jobs:
            dockerfile: "./backend/Dockerfile.python"
            context: "./"
            ubuntu-version: '2404'
-          - build-type: ''
-            cuda-major-version: ""
-            cuda-minor-version: ""
-            platforms: 'linux/amd64,linux/arm64'
-            tag-latest: 'auto'
-            tag-suffix: '-cpu-qwen-tts'
-            runs-on: 'ubuntu-latest'
-            base-image: "ubuntu:24.04"
-            skip-drivers: 'false'
-            backend: "qwen-tts"
-            dockerfile: "./backend/Dockerfile.python"
-            context: "./"
-            ubuntu-version: '2404'
-          - build-type: ''
-            cuda-major-version: ""
-            cuda-minor-version: ""
-            platforms: 'linux/amd64,linux/arm64'
-            tag-latest: 'auto'
-            tag-suffix: '-cpu-pocket-tts'
-            runs-on: 'ubuntu-latest'
-            base-image: "ubuntu:24.04"
-            skip-drivers: 'false'
-            backend: "pocket-tts"
-            dockerfile: "./backend/Dockerfile.python"
-            context: "./"
-            ubuntu-version: '2404'
  backend-jobs-darwin:
    uses: ./.github/workflows/backend_build_darwin.yml
    strategy:
--- a/.github/workflows/test-extra.yml
+++ b/.github/workflows/test-extra.yml
@@ -265,42 +265,4 @@ jobs:
      - name: Test moonshine
        run: |
          make --jobs=5 --output-sync=target -C backend/python/moonshine
-          make --jobs=5 --output-sync=target -C backend/python/moonshine test
-  tests-pocket-tts:
-    runs-on: ubuntu-latest
-    steps:
-      - name: Clone
-        uses: actions/checkout@v6
-        with:
-          submodules: true
-      - name: Dependencies
-        run: |
-          sudo apt-get update
-          sudo apt-get install build-essential ffmpeg
-          sudo apt-get install -y ca-certificates cmake curl patch python3-pip
-          # Install UV
-          curl -LsSf https://astral.sh/uv/install.sh | sh
-          pip install --user --no-cache-dir grpcio-tools==1.64.1
-      - name: Test pocket-tts
-        run: |
-          make --jobs=5 --output-sync=target -C backend/python/pocket-tts
-          make --jobs=5 --output-sync=target -C backend/python/pocket-tts test
-  tests-qwen-tts:
-    runs-on: ubuntu-latest
-    steps:
-      - name: Clone
-        uses: actions/checkout@v6
-        with:
-          submodules: true
-      - name: Dependencies
-        run: |
-          sudo apt-get update
-          sudo apt-get install build-essential ffmpeg
-          sudo apt-get install -y ca-certificates cmake curl patch python3-pip
-          # Install UV
-          curl -LsSf https://astral.sh/uv/install.sh | sh
-          pip install --user --no-cache-dir grpcio-tools==1.64.1
-      - name: Test qwen-tts
-        run: |
-          make --jobs=5 --output-sync=target -C backend/python/qwen-tts
-          make --jobs=5 --output-sync=target -C backend/python/qwen-tts test
+          make --jobs=5 --output-sync=target -C backend/python/moonshine test
--- a/20
+++ b/20
@@ -10,7 +10,7 @@ ENV DEBIAN_FRONTEND=noninteractive
 RUN apt-get update && \
    apt-get install -y --no-install-recommends \
        ca-certificates curl wget espeak-ng libgomp1 \
-        ffmpeg libopenblas0 libopenblas-dev sox && \
+        ffmpeg libopenblas0 libopenblas-dev && \
    apt-get clean && \
    rm -rf /var/lib/apt/lists/*

@@ -42,22 +42,22 @@ RUN <<EOT bash
            ocaml-core ninja-build pkg-config libxml2-dev wayland-protocols python3-jsonschema \
            clang-format qtbase5-dev qt6-base-dev libxcb-glx0-dev sudo xz-utils mesa-vulkan-drivers
        if [ "amd64" = "$TARGETARCH" ]; then
-            wget "https://sdk.lunarg.com/sdk/download/1.4.335.0/linux/vulkansdk-linux-x86_64-1.4.335.0.tar.xz" && \
-            tar -xf vulkansdk-linux-x86_64-1.4.335.0.tar.xz && \
-            rm vulkansdk-linux-x86_64-1.4.335.0.tar.xz && \
+            wget "https://sdk.lunarg.com/sdk/download/1.4.328.1/linux/vulkansdk-linux-x86_64-1.4.328.1.tar.xz" && \
+            tar -xf vulkansdk-linux-x86_64-1.4.328.1.tar.xz && \
+            rm vulkansdk-linux-x86_64-1.4.328.1.tar.xz && \
            mkdir -p /opt/vulkan-sdk && \
-            mv 1.4.335.0 /opt/vulkan-sdk/ && \
-            cd /opt/vulkan-sdk/1.4.335.0 && \
+            mv 1.4.328.1 /opt/vulkan-sdk/ && \
+            cd /opt/vulkan-sdk/1.4.328.1 && \
            ./vulkansdk --no-deps --maxjobs \
                vulkan-loader \
                vulkan-validationlayers \
                vulkan-extensionlayer \
                vulkan-tools \
                shaderc && \
-            cp -rfv /opt/vulkan-sdk/1.4.335.0/x86_64/bin/* /usr/bin/ && \
-            cp -rfv /opt/vulkan-sdk/1.4.335.0/x86_64/lib/* /usr/lib/x86_64-linux-gnu/ && \
-            cp -rfv /opt/vulkan-sdk/1.4.335.0/x86_64/include/* /usr/include/ && \
-            cp -rfv /opt/vulkan-sdk/1.4.335.0/x86_64/share/* /usr/share/ && \
+            cp -rfv /opt/vulkan-sdk/1.4.328.1/x86_64/bin/* /usr/bin/ && \
+            cp -rfv /opt/vulkan-sdk/1.4.328.1/x86_64/lib/* /usr/lib/x86_64-linux-gnu/ && \
+            cp -rfv /opt/vulkan-sdk/1.4.328.1/x86_64/include/* /usr/include/ && \
+            cp -rfv /opt/vulkan-sdk/1.4.328.1/x86_64/share/* /usr/share/ && \
            rm -rf /opt/vulkan-sdk
        fi
        if [ "arm64" = "$TARGETARCH" ]; then
--- a/34
+++ b/34
@@ -1,5 +1,5 @@
 # Disable parallel execution for backend builds
-.NOTPARALLEL: backends/diffusers backends/llama-cpp backends/piper backends/stablediffusion-ggml backends/whisper backends/faster-whisper backends/silero-vad backends/local-store backends/huggingface backends/rfdetr backends/kitten-tts backends/kokoro backends/chatterbox backends/llama-cpp-darwin backends/neutts build-darwin-python-backend build-darwin-go-backend backends/mlx backends/diffuser-darwin backends/mlx-vlm backends/mlx-audio backends/stablediffusion-ggml-darwin backends/vllm backends/moonshine backends/pocket-tts backends/qwen-tts
+.NOTPARALLEL: backends/diffusers backends/llama-cpp backends/piper backends/stablediffusion-ggml backends/whisper backends/faster-whisper backends/silero-vad backends/local-store backends/huggingface backends/rfdetr backends/kitten-tts backends/kokoro backends/chatterbox backends/llama-cpp-darwin backends/neutts build-darwin-python-backend build-darwin-go-backend backends/mlx backends/diffuser-darwin backends/mlx-vlm backends/mlx-audio backends/stablediffusion-ggml-darwin backends/vllm backends/moonshine

 GOCMD=go
 GOTEST=$(GOCMD) test
@@ -9,7 +9,7 @@ LAUNCHER_BINARY_NAME=local-ai-launcher

 CUDA_MAJOR_VERSION?=13
 CUDA_MINOR_VERSION?=0
-UBUNTU_VERSION?=2404
+UBUNTU_VERSION?=2204
 UBUNTU_CODENAME?=noble

 GORELEASER?=
@@ -316,8 +316,6 @@ prepare-test-extra: protogen-python
 	$(MAKE) -C backend/python/vllm
 	$(MAKE) -C backend/python/vibevoice
 	$(MAKE) -C backend/python/moonshine
-	$(MAKE) -C backend/python/pocket-tts
-	$(MAKE) -C backend/python/qwen-tts

 test-extra: prepare-test-extra
 	$(MAKE) -C backend/python/transformers test
@@ -326,8 +324,6 @@ test-extra: prepare-test-extra
 	$(MAKE) -C backend/python/vllm test
 	$(MAKE) -C backend/python/vibevoice test
 	$(MAKE) -C backend/python/moonshine test
-	$(MAKE) -C backend/python/pocket-tts test
-	$(MAKE) -C backend/python/qwen-tts test

 DOCKER_IMAGE?=local-ai
 DOCKER_AIO_IMAGE?=local-ai-aio
@@ -451,17 +447,17 @@ BACKEND_FASTER_WHISPER = faster-whisper|python|.|false|true
 BACKEND_COQUI = coqui|python|.|false|true
 BACKEND_BARK = bark|python|.|false|true
 BACKEND_EXLLAMA2 = exllama2|python|.|false|true
-BACKEND_RFDETR = rfdetr|python|.|false|true
-BACKEND_KITTEN_TTS = kitten-tts|python|.|false|true
-BACKEND_NEUTTS = neutts|python|.|false|true
-BACKEND_KOKORO = kokoro|python|.|false|true
-BACKEND_VLLM = vllm|python|.|false|true
-BACKEND_DIFFUSERS = diffusers|python|.|--progress=plain|true
-BACKEND_CHATTERBOX = chatterbox|python|.|false|true
-BACKEND_VIBEVOICE = vibevoice|python|.|--progress=plain|true
-BACKEND_MOONSHINE = moonshine|python|.|false|true
-BACKEND_POCKET_TTS = pocket-tts|python|.|false|true
-BACKEND_QWEN_TTS = qwen-tts|python|.|false|true
+
+# Python backends with ./backend context
+BACKEND_RFDETR = rfdetr|python|./backend|false|true
+BACKEND_KITTEN_TTS = kitten-tts|python|./backend|false|true
+BACKEND_NEUTTS = neutts|python|./backend|false|true
+BACKEND_KOKORO = kokoro|python|./backend|false|true
+BACKEND_VLLM = vllm|python|./backend|false|true
+BACKEND_DIFFUSERS = diffusers|python|./backend|--progress=plain|true
+BACKEND_CHATTERBOX = chatterbox|python|./backend|false|true
+BACKEND_VIBEVOICE = vibevoice|python|./backend|--progress=plain|true
+BACKEND_MOONSHINE = moonshine|python|./backend|false|true

 # Helper function to build docker image for a backend
 # Usage: $(call docker-build-backend,BACKEND_NAME,DOCKERFILE_TYPE,BUILD_CONTEXT,PROGRESS_FLAG,NEEDS_BACKEND_ARG)
@@ -507,14 +503,12 @@ $(eval $(call generate-docker-build-target,$(BACKEND_DIFFUSERS)))
 $(eval $(call generate-docker-build-target,$(BACKEND_CHATTERBOX)))
 $(eval $(call generate-docker-build-target,$(BACKEND_VIBEVOICE)))
 $(eval $(call generate-docker-build-target,$(BACKEND_MOONSHINE)))
-$(eval $(call generate-docker-build-target,$(BACKEND_POCKET_TTS)))
-$(eval $(call generate-docker-build-target,$(BACKEND_QWEN_TTS)))

 # Pattern rule for docker-save targets
 docker-save-%: backend-images
 	docker save local-ai-backend:$* -o backend-images/$*.tar

-docker-build-backends: docker-build-llama-cpp docker-build-rerankers docker-build-vllm docker-build-transformers docker-build-diffusers docker-build-kokoro docker-build-faster-whisper docker-build-coqui docker-build-bark docker-build-chatterbox docker-build-vibevoice docker-build-exllama2 docker-build-moonshine docker-build-pocket-tts docker-build-qwen-tts
+docker-build-backends: docker-build-llama-cpp docker-build-rerankers docker-build-vllm docker-build-transformers docker-build-diffusers docker-build-kokoro docker-build-faster-whisper docker-build-coqui docker-build-bark docker-build-chatterbox docker-build-vibevoice docker-build-exllama2 docker-build-moonshine

 ########################################################
 ### END Backends
--- a/README.md
+++ b/README.md
@@ -111,8 +111,6 @@

 ## 💻 Quickstart

-> ⚠️ **Note:** The `install.sh` script is currently experiencing issues due to the heavy changes currently undergoing in LocalAI and may produce broken or misconfigured installations. Please use Docker installation (see below) or manual binary installation until [issue #8032](https://github.com/mudler/LocalAI/issues/8032) is resolved.
-
 Run the installer script:

 ```bash
@@ -130,7 +128,7 @@ For more installation options, see [Installer Options](https://localai.io/instal

 > Note: the DMGs are not signed by Apple as quarantined. See https://github.com/mudler/LocalAI/issues/6268 for a workaround, fix is tracked here: https://github.com/mudler/LocalAI/issues/6244

-### Containers (Docker, podman, ...)
+Or run with docker:

 > **💡 Docker Run vs Docker Start**
 > 
@@ -139,13 +137,13 @@ For more installation options, see [Installer Options](https://localai.io/instal
 > 
 > If you've already run LocalAI before and want to start it again, use: `docker start -i local-ai`

-#### CPU only image:
+### CPU only image:

 ```bash
 docker run -ti --name local-ai -p 8080:8080 localai/localai:latest
 ```

-#### NVIDIA GPU Images:
+### NVIDIA GPU Images:

 ```bash
 # CUDA 13.0
@@ -162,25 +160,25 @@ docker run -ti --name local-ai -p 8080:8080 --gpus all localai/localai:latest-nv
 docker run -ti --name local-ai -p 8080:8080 --gpus all localai/localai:latest-nvidia-l4t-arm64-cuda-13
 ```

-#### AMD GPU Images (ROCm):
+### AMD GPU Images (ROCm):

 ```bash
 docker run -ti --name local-ai -p 8080:8080 --device=/dev/kfd --device=/dev/dri --group-add=video localai/localai:latest-gpu-hipblas
 ```

-#### Intel GPU Images (oneAPI):
+### Intel GPU Images (oneAPI):

 ```bash
 docker run -ti --name local-ai -p 8080:8080 --device=/dev/dri/card1 --device=/dev/dri/renderD128 localai/localai:latest-gpu-intel
 ```

-#### Vulkan GPU Images:
+### Vulkan GPU Images:

 ```bash
 docker run -ti --name local-ai -p 8080:8080 localai/localai:latest-gpu-vulkan
 ```

-#### AIO Images (pre-downloaded models):
+### AIO Images (pre-downloaded models):

 ```bash
 # CPU version
@@ -297,8 +295,6 @@ LocalAI supports a comprehensive range of AI backends with multiple acceleration
 | **silero-vad** | Voice Activity Detection | CPU |
 | **neutts** | Text-to-speech with voice cloning | CUDA 12/13, ROCm, CPU |
 | **vibevoice** | Real-time TTS with voice cloning | CUDA 12/13, ROCm, Intel, CPU |
-| **pocket-tts** | Lightweight CPU-based TTS | CUDA 12/13, ROCm, Intel, CPU |
-| **qwen-tts** | High-quality TTS with custom voice, voice design, and voice cloning | CUDA 12/13, ROCm, Intel, CPU |

 ### Image & Video Generation
 | Backend | Description | Acceleration Support |
@@ -320,8 +316,8 @@ LocalAI supports a comprehensive range of AI backends with multiple acceleration
 |-------------------|-------------------|------------------|
 | **NVIDIA CUDA 12** | All CUDA-compatible backends | Nvidia hardware |
 | **NVIDIA CUDA 13** | All CUDA-compatible backends | Nvidia hardware |
-| **AMD ROCm** | llama.cpp, whisper, vllm, transformers, diffusers, rerankers, coqui, kokoro, bark, neutts, vibevoice, pocket-tts, qwen-tts | AMD Graphics |
-| **Intel oneAPI** | llama.cpp, whisper, stablediffusion, vllm, transformers, diffusers, rfdetr, rerankers, exllama2, coqui, kokoro, bark, vibevoice, pocket-tts, qwen-tts | Intel Arc, Intel iGPUs |
+| **AMD ROCm** | llama.cpp, whisper, vllm, transformers, diffusers, rerankers, coqui, kokoro, bark, neutts, vibevoice | AMD Graphics |
+| **Intel oneAPI** | llama.cpp, whisper, stablediffusion, vllm, transformers, diffusers, rfdetr, rerankers, exllama2, coqui, kokoro, bark, vibevoice | Intel Arc, Intel iGPUs |
 | **Apple Metal** | llama.cpp, whisper, diffusers, MLX, MLX-VLM, bark-cpp | Apple M1/M2/M3+ |
 | **Vulkan** | llama.cpp, whisper, stablediffusion | Cross-platform GPUs |
 | **NVIDIA Jetson (CUDA 12)** | llama.cpp, whisper, stablediffusion, diffusers, rfdetr | ARM64 embedded AI (AGX Orin, etc.) |
--- a/backend/Dockerfile.golang
+++ b/backend/Dockerfile.golang
@@ -47,22 +47,22 @@ RUN <<EOT bash
            ocaml-core ninja-build pkg-config libxml2-dev wayland-protocols python3-jsonschema \
            clang-format qtbase5-dev qt6-base-dev libxcb-glx0-dev sudo xz-utils
        if [ "amd64" = "$TARGETARCH" ]; then
-            wget "https://sdk.lunarg.com/sdk/download/1.4.335.0/linux/vulkansdk-linux-x86_64-1.4.335.0.tar.xz" && \
-            tar -xf vulkansdk-linux-x86_64-1.4.335.0.tar.xz && \
-            rm vulkansdk-linux-x86_64-1.4.335.0.tar.xz && \
+            wget "https://sdk.lunarg.com/sdk/download/1.4.328.1/linux/vulkansdk-linux-x86_64-1.4.328.1.tar.xz" && \
+            tar -xf vulkansdk-linux-x86_64-1.4.328.1.tar.xz && \
+            rm vulkansdk-linux-x86_64-1.4.328.1.tar.xz && \
            mkdir -p /opt/vulkan-sdk && \
-            mv 1.4.335.0 /opt/vulkan-sdk/ && \
-            cd /opt/vulkan-sdk/1.4.335.0 && \
+            mv 1.4.328.1 /opt/vulkan-sdk/ && \
+            cd /opt/vulkan-sdk/1.4.328.1 && \
            ./vulkansdk --no-deps --maxjobs \
                vulkan-loader \
                vulkan-validationlayers \
                vulkan-extensionlayer \
                vulkan-tools \
                shaderc && \
-            cp -rfv /opt/vulkan-sdk/1.4.335.0/x86_64/bin/* /usr/bin/ && \
-            cp -rfv /opt/vulkan-sdk/1.4.335.0/x86_64/lib/* /usr/lib/x86_64-linux-gnu/ && \
-            cp -rfv /opt/vulkan-sdk/1.4.335.0/x86_64/include/* /usr/include/ && \
-            cp -rfv /opt/vulkan-sdk/1.4.335.0/x86_64/share/* /usr/share/ && \
+            cp -rfv /opt/vulkan-sdk/1.4.328.1/x86_64/bin/* /usr/bin/ && \
+            cp -rfv /opt/vulkan-sdk/1.4.328.1/x86_64/lib/* /usr/lib/x86_64-linux-gnu/ && \
+            cp -rfv /opt/vulkan-sdk/1.4.328.1/x86_64/include/* /usr/include/ && \
+            cp -rfv /opt/vulkan-sdk/1.4.328.1/x86_64/share/* /usr/share/ && \
            rm -rf /opt/vulkan-sdk
        fi
        if [ "arm64" = "$TARGETARCH" ]; then
--- a/backend/Dockerfile.llama-cpp
+++ b/backend/Dockerfile.llama-cpp
@@ -104,22 +104,22 @@ RUN <<EOT bash
            ocaml-core ninja-build pkg-config libxml2-dev wayland-protocols python3-jsonschema \
            clang-format qtbase5-dev qt6-base-dev libxcb-glx0-dev sudo xz-utils
        if [ "amd64" = "$TARGETARCH" ]; then
-            wget "https://sdk.lunarg.com/sdk/download/1.4.335.0/linux/vulkansdk-linux-x86_64-1.4.335.0.tar.xz" && \
-            tar -xf vulkansdk-linux-x86_64-1.4.335.0.tar.xz && \
-            rm vulkansdk-linux-x86_64-1.4.335.0.tar.xz && \
+            wget "https://sdk.lunarg.com/sdk/download/1.4.328.1/linux/vulkansdk-linux-x86_64-1.4.328.1.tar.xz" && \
+            tar -xf vulkansdk-linux-x86_64-1.4.328.1.tar.xz && \
+            rm vulkansdk-linux-x86_64-1.4.328.1.tar.xz && \
            mkdir -p /opt/vulkan-sdk && \
-            mv 1.4.335.0 /opt/vulkan-sdk/ && \
-            cd /opt/vulkan-sdk/1.4.335.0 && \
+            mv 1.4.328.1 /opt/vulkan-sdk/ && \
+            cd /opt/vulkan-sdk/1.4.328.1 && \
            ./vulkansdk --no-deps --maxjobs \
                vulkan-loader \
                vulkan-validationlayers \
                vulkan-extensionlayer \
                vulkan-tools \
                shaderc && \
-            cp -rfv /opt/vulkan-sdk/1.4.335.0/x86_64/bin/* /usr/bin/ && \
-            cp -rfv /opt/vulkan-sdk/1.4.335.0/x86_64/lib/* /usr/lib/x86_64-linux-gnu/ && \
-            cp -rfv /opt/vulkan-sdk/1.4.335.0/x86_64/include/* /usr/include/ && \
-            cp -rfv /opt/vulkan-sdk/1.4.335.0/x86_64/share/* /usr/share/ && \
+            cp -rfv /opt/vulkan-sdk/1.4.328.1/x86_64/bin/* /usr/bin/ && \
+            cp -rfv /opt/vulkan-sdk/1.4.328.1/x86_64/lib/* /usr/lib/x86_64-linux-gnu/ && \
+            cp -rfv /opt/vulkan-sdk/1.4.328.1/x86_64/include/* /usr/include/ && \
+            cp -rfv /opt/vulkan-sdk/1.4.328.1/x86_64/share/* /usr/share/ && \
            rm -rf /opt/vulkan-sdk
        fi
        if [ "arm64" = "$TARGETARCH" ]; then
--- a/backend/Dockerfile.python
+++ b/backend/Dockerfile.python
@@ -61,22 +61,22 @@ RUN <<EOT bash
            ocaml-core ninja-build pkg-config libxml2-dev wayland-protocols python3-jsonschema \
            clang-format qtbase5-dev qt6-base-dev libxcb-glx0-dev sudo xz-utils
        if [ "amd64" = "$TARGETARCH" ]; then
-            wget "https://sdk.lunarg.com/sdk/download/1.4.335.0/linux/vulkansdk-linux-x86_64-1.4.335.0.tar.xz" && \
-            tar -xf vulkansdk-linux-x86_64-1.4.335.0.tar.xz && \
-            rm vulkansdk-linux-x86_64-1.4.335.0.tar.xz && \
+            wget "https://sdk.lunarg.com/sdk/download/1.4.328.1/linux/vulkansdk-linux-x86_64-1.4.328.1.tar.xz" && \
+            tar -xf vulkansdk-linux-x86_64-1.4.328.1.tar.xz && \
+            rm vulkansdk-linux-x86_64-1.4.328.1.tar.xz && \
            mkdir -p /opt/vulkan-sdk && \
-            mv 1.4.335.0 /opt/vulkan-sdk/ && \
-            cd /opt/vulkan-sdk/1.4.335.0 && \
+            mv 1.4.328.1 /opt/vulkan-sdk/ && \
+            cd /opt/vulkan-sdk/1.4.328.1 && \
            ./vulkansdk --no-deps --maxjobs \
                vulkan-loader \
                vulkan-validationlayers \
                vulkan-extensionlayer \
                vulkan-tools \
                shaderc && \
-            cp -rfv /opt/vulkan-sdk/1.4.335.0/x86_64/bin/* /usr/bin/ && \
-            cp -rfv /opt/vulkan-sdk/1.4.335.0/x86_64/lib/* /usr/lib/x86_64-linux-gnu/ && \
-            cp -rfv /opt/vulkan-sdk/1.4.335.0/x86_64/include/* /usr/include/ && \
-            cp -rfv /opt/vulkan-sdk/1.4.335.0/x86_64/share/* /usr/share/ && \
+            cp -rfv /opt/vulkan-sdk/1.4.328.1/x86_64/bin/* /usr/bin/ && \
+            cp -rfv /opt/vulkan-sdk/1.4.328.1/x86_64/lib/* /usr/lib/x86_64-linux-gnu/ && \
+            cp -rfv /opt/vulkan-sdk/1.4.328.1/x86_64/include/* /usr/include/ && \
+            cp -rfv /opt/vulkan-sdk/1.4.328.1/x86_64/share/* /usr/share/ && \
            rm -rf /opt/vulkan-sdk
        fi
        if [ "arm64" = "$TARGETARCH" ]; then
--- a/backend/backend.proto
+++ b/backend/backend.proto
@@ -32,8 +32,6 @@ service Backend {
  rpc GetMetrics(MetricsRequest) returns (MetricsResponse);

  rpc VAD(VADRequest) returns (VADResponse) {}
-
-  rpc ModelMetadata(ModelOptions) returns (ModelMetadataResponse) {}
 }

 // Define the empty request
@@ -412,8 +410,3 @@ message Detection {
 message DetectResponse {
  repeated Detection Detections = 1;
 }
-
-message ModelMetadataResponse {
-  bool supports_thinking = 1;
-  string rendered_template = 2;  // The rendered chat template with enable_thinking=true (empty if not applicable)
-}
--- a/backend/cpp/llama-cpp/Makefile
+++ b/backend/cpp/llama-cpp/Makefile
@@ -1,5 +1,5 @@

-LLAMA_VERSION?=a5eaa1d6a3732bc0f460b02b61c95680bba5a012
+LLAMA_VERSION?=b1377188784f9aea26b8abde56d4aee8c733eec7
 LLAMA_REPO?=https://github.com/ggerganov/llama.cpp

 CMAKE_ARGS?=
--- a/backend/cpp/llama-cpp/grpc-server.cpp
+++ b/backend/cpp/llama-cpp/grpc-server.cpp
@@ -83,8 +83,8 @@ static void start_llama_server(server_context& ctx_server) {

    // print sample chat example to make it clear which template is used
    // LOG_INF("%s: chat template, chat_template: %s, example_format: '%s'\n", __func__,
-    //     common_chat_templates_source(ctx_server.impl->chat_params.tmpls.get()),
-    //     common_chat_format_example(ctx_server.impl->chat_params.tmpls.get(), ctx_server.impl->params_base.use_jinja).c_str(), ctx_server.impl->params_base.default_template_kwargs);
+    //     common_chat_templates_source(ctx_server.impl->chat_templates.get()),
+    //     common_chat_format_example(ctx_server.impl->chat_templates.get(), ctx_server.impl->params_base.use_jinja).c_str(), ctx_server.impl->params_base.default_template_kwargs);

    // Keep the chat templates initialized in load_model() so they can be used when UseTokenizerTemplate is enabled
    // Templates will only be used conditionally in Predict/PredictStream when UseTokenizerTemplate is true and Messages are provided
@@ -882,7 +882,7 @@ public:
            std::string prompt_str;
            std::vector<raw_buffer> files; // Declare files early so it's accessible in both branches
            // Handle chat templates when UseTokenizerTemplate is enabled and Messages are provided
-            if (request->usetokenizertemplate() && request->messages_size() > 0 && ctx_server.impl->chat_params.tmpls != nullptr) {
+            if (request->usetokenizertemplate() && request->messages_size() > 0 && ctx_server.impl->chat_templates != nullptr) {
                // Convert proto Messages to JSON format compatible with oaicompat_chat_params_parse
                json body_json;
                json messages_json = json::array();
@@ -1261,7 +1261,12 @@ public:
                // Use the same approach as server.cpp: call oaicompat_chat_params_parse
                // This handles all template application, grammar merging, etc. automatically
                // Files extracted from multimodal content in messages will be added to the files vector
-                // chat_params already contains tmpls, allow_image, and allow_audio set during model loading
+                // Create parser options with current chat_templates to ensure tmpls is not null
+                oaicompat_parser_options parser_opt = ctx_server.impl->oai_parser_opt;
+                parser_opt.tmpls = ctx_server.impl->chat_templates.get(); // Ensure tmpls is set to current chat_templates
+                // Update allow_image and allow_audio based on current mctx state
+                parser_opt.allow_image = ctx_server.impl->mctx ? mtmd_support_vision(ctx_server.impl->mctx) : false;
+                parser_opt.allow_audio = ctx_server.impl->mctx ? mtmd_support_audio(ctx_server.impl->mctx) : false;

                // Debug: Log tools before template processing
                if (body_json.contains("tools")) {
@@ -1307,7 +1312,7 @@ public:
                    }
                }

-                json parsed_data = oaicompat_chat_params_parse(body_json, ctx_server.impl->chat_params, files);
+                json parsed_data = oaicompat_chat_params_parse(body_json, parser_opt, files);

                // Debug: Log tools after template processing
                if (parsed_data.contains("tools")) {
@@ -1360,7 +1365,7 @@ public:

            // If not using chat templates, extract files from image_data/audio_data fields
            // (If using chat templates, files were already extracted by oaicompat_chat_params_parse)
-            if (!request->usetokenizertemplate() || request->messages_size() == 0 || ctx_server.impl->chat_params.tmpls == nullptr) {
+            if (!request->usetokenizertemplate() || request->messages_size() == 0 || ctx_server.impl->chat_templates == nullptr) {
                const auto &images_data = data.find("image_data");
                if (images_data != data.end() && images_data->is_array())
                {
@@ -1588,7 +1593,7 @@ public:
            std::string prompt_str;
            std::vector<raw_buffer> files; // Declare files early so it's accessible in both branches
            // Handle chat templates when UseTokenizerTemplate is enabled and Messages are provided
-            if (request->usetokenizertemplate() && request->messages_size() > 0 && ctx_server.impl->chat_params.tmpls != nullptr) {
+            if (request->usetokenizertemplate() && request->messages_size() > 0 && ctx_server.impl->chat_templates != nullptr) {
                // Convert proto Messages to JSON format compatible with oaicompat_chat_params_parse
                json body_json;
                json messages_json = json::array();
@@ -1992,7 +1997,12 @@ public:
                // Use the same approach as server.cpp: call oaicompat_chat_params_parse
                // This handles all template application, grammar merging, etc. automatically
                // Files extracted from multimodal content in messages will be added to the files vector
-                // chat_params already contains tmpls, allow_image, and allow_audio set during model loading
+                // Create parser options with current chat_templates to ensure tmpls is not null
+                oaicompat_parser_options parser_opt = ctx_server.impl->oai_parser_opt;
+                parser_opt.tmpls = ctx_server.impl->chat_templates.get(); // Ensure tmpls is set to current chat_templates
+                // Update allow_image and allow_audio based on current mctx state
+                parser_opt.allow_image = ctx_server.impl->mctx ? mtmd_support_vision(ctx_server.impl->mctx) : false;
+                parser_opt.allow_audio = ctx_server.impl->mctx ? mtmd_support_audio(ctx_server.impl->mctx) : false;

                // Debug: Log tools before template processing
                if (body_json.contains("tools")) {
@@ -2038,7 +2048,7 @@ public:
                    }
                }

-                json parsed_data = oaicompat_chat_params_parse(body_json, ctx_server.impl->chat_params, files);
+                json parsed_data = oaicompat_chat_params_parse(body_json, parser_opt, files);

                // Debug: Log tools after template processing
                if (parsed_data.contains("tools")) {
@@ -2091,7 +2101,7 @@ public:

            // If not using chat templates, extract files from image_data/audio_data fields
            // (If using chat templates, files were already extracted by oaicompat_chat_params_parse)
-            if (!request->usetokenizertemplate() || request->messages_size() == 0 || ctx_server.impl->chat_params.tmpls == nullptr) {
+            if (!request->usetokenizertemplate() || request->messages_size() == 0 || ctx_server.impl->chat_templates == nullptr) {
                const auto &images_data = data.find("image_data");
                if (images_data != data.end() && images_data->is_array())
                {
@@ -2476,47 +2486,6 @@ public:
        response->set_prompt_tokens_processed(res_metrics->n_prompt_tokens_processed_total);


-        return grpc::Status::OK;
-    }
-
-    grpc::Status ModelMetadata(ServerContext* /*context*/, const backend::ModelOptions* /*request*/, backend::ModelMetadataResponse* response) override {
-        // Check if model is loaded
-        if (params_base.model.path.empty()) {
-            return grpc::Status(grpc::StatusCode::FAILED_PRECONDITION, "Model not loaded");
-        }
-
-        // Check if chat templates are initialized
-        if (ctx_server.impl->chat_params.tmpls == nullptr) {
-            // If templates are not initialized, we can't detect thinking support
-            // Return false as default
-            response->set_supports_thinking(false);
-            response->set_rendered_template("");
-            return grpc::Status::OK;
-        }
-
-        // Detect thinking support using llama.cpp's function
-        bool supports_thinking = common_chat_templates_support_enable_thinking(ctx_server.impl->chat_params.tmpls.get());
-        response->set_supports_thinking(supports_thinking);
-
-        // Render the template with enable_thinking=true so Go code can detect thinking tokens
-        // This allows reusing existing detection functions in Go
-        std::string rendered_template = "";
-        if (params_base.use_jinja) {
-            // Render the template with enable_thinking=true to see what the actual prompt looks like
-            common_chat_templates_inputs dummy_inputs;
-            common_chat_msg msg;
-            msg.role = "user";
-            msg.content = "test";
-            dummy_inputs.messages = {msg};
-            dummy_inputs.enable_thinking = true;
-            dummy_inputs.use_jinja = params_base.use_jinja;
-            
-            const auto rendered = common_chat_templates_apply(ctx_server.impl->chat_params.tmpls.get(), dummy_inputs);
-            rendered_template = rendered.prompt;
-        }
-        
-        response->set_rendered_template(rendered_template);
-
        return grpc::Status::OK;
    }
 };
--- a/backend/go/stablediffusion-ggml/Makefile
+++ b/backend/go/stablediffusion-ggml/Makefile
@@ -8,7 +8,7 @@ JOBS?=$(shell nproc --ignore=1)

 # stablediffusion.cpp (ggml)
 STABLEDIFFUSION_GGML_REPO?=https://github.com/leejet/stable-diffusion.cpp
-STABLEDIFFUSION_GGML_VERSION?=5e4579c11d0678f9765463582d024e58270faa9c
+STABLEDIFFUSION_GGML_VERSION?=0e52afc6513cc2dea9a1a017afc4a008d5acf2b0

 CMAKE_ARGS+=-DGGML_MAX_NAME=128

--- a/backend/go/whisper/Makefile
+++ b/backend/go/whisper/Makefile
@@ -8,7 +8,7 @@ JOBS?=$(shell nproc --ignore=1)

 # whisper.cpp version
 WHISPER_REPO?=https://github.com/ggml-org/whisper.cpp
-WHISPER_CPP_VERSION?=7aa8818647303b567c3a21fe4220b2681988e220
+WHISPER_CPP_VERSION?=679bdb53dbcbfb3e42685f50c7ff367949fd4d48
 SO_TARGET?=libgowhisper.so

 CMAKE_ARGS+=-DBUILD_SHARED_LIBS=OFF
--- a/backend/index.yaml
+++ b/backend/index.yaml
@@ -428,50 +428,6 @@
    nvidia-l4t-cuda-12: "nvidia-l4t-vibevoice"
    nvidia-l4t-cuda-13: "cuda13-nvidia-l4t-arm64-vibevoice"
  icon: https://avatars.githubusercontent.com/u/6154722?s=200&v=4
- &qwen-tts
-  urls:
-    - https://github.com/QwenLM/Qwen3-TTS
-  description: |
-    Qwen3-TTS is a high-quality text-to-speech model supporting custom voice, voice design, and voice cloning.
-  tags:
-    - text-to-speech
-    - TTS
-  license: apache-2.0
-  name: "qwen-tts"
-  alias: "qwen-tts"
-  capabilities:
-    nvidia: "cuda12-qwen-tts"
-    intel: "intel-qwen-tts"
-    amd: "rocm-qwen-tts"
-    nvidia-l4t: "nvidia-l4t-qwen-tts"
-    default: "cpu-qwen-tts"
-    nvidia-cuda-13: "cuda13-qwen-tts"
-    nvidia-cuda-12: "cuda12-qwen-tts"
-    nvidia-l4t-cuda-12: "nvidia-l4t-qwen-tts"
-    nvidia-l4t-cuda-13: "cuda13-nvidia-l4t-arm64-qwen-tts"
-  icon: https://avatars.githubusercontent.com/u/6154722?s=200&v=4
- &pocket-tts
-  urls:
-    - https://github.com/kyutai-labs/pocket-tts
-  description: |
-    Pocket TTS is a lightweight text-to-speech model designed to run efficiently on CPUs.
-  tags:
-    - text-to-speech
-    - TTS
-  license: mit
-  name: "pocket-tts"
-  alias: "pocket-tts"
-  capabilities:
-    nvidia: "cuda12-pocket-tts"
-    intel: "intel-pocket-tts"
-    amd: "rocm-pocket-tts"
-    nvidia-l4t: "nvidia-l4t-pocket-tts"
-    default: "cpu-pocket-tts"
-    nvidia-cuda-13: "cuda13-pocket-tts"
-    nvidia-cuda-12: "cuda12-pocket-tts"
-    nvidia-l4t-cuda-12: "nvidia-l4t-pocket-tts"
-    nvidia-l4t-cuda-13: "cuda13-nvidia-l4t-arm64-pocket-tts"
-  icon: https://avatars.githubusercontent.com/u/6154722?s=200&v=4
 - &piper
  name: "piper"
  uri: "quay.io/go-skynet/local-ai-backends:latest-piper"
@@ -559,14 +515,18 @@
    default: "cpu-neutts"
    nvidia: "cuda12-neutts"
    amd: "rocm-neutts"
+    nvidia-l4t: "nvidia-l4t-neutts"
    nvidia-cuda-12: "cuda12-neutts"
+    nvidia-l4t-cuda-12: "nvidia-l4t-arm64-neutts"
 - !!merge <<: *neutts
  name: "neutts-development"
  capabilities:
    default: "cpu-neutts-development"
    nvidia: "cuda12-neutts-development"
    amd: "rocm-neutts-development"
+    nvidia-l4t: "nvidia-l4t-neutts-development"
    nvidia-cuda-12: "cuda12-neutts-development"
+    nvidia-l4t-cuda-12: "nvidia-l4t-arm64-neutts-development"
 - !!merge <<: *llamacpp
  name: "llama-cpp-development"
  capabilities:
@@ -596,6 +556,11 @@
  uri: "quay.io/go-skynet/local-ai-backends:latest-gpu-rocm-hipblas-neutts"
  mirrors:
    - localai/localai-backends:latest-gpu-rocm-hipblas-neutts
+- !!merge <<: *neutts
+  name: "nvidia-l4t-arm64-neutts"
+  uri: "quay.io/go-skynet/local-ai-backends:latest-nvidia-l4t-arm64-neutts"
+  mirrors:
+    - localai/localai-backends:latest-nvidia-l4t-arm64-neutts
 - !!merge <<: *neutts
  name: "cpu-neutts-development"
  uri: "quay.io/go-skynet/local-ai-backends:master-cpu-neutts"
@@ -611,6 +576,11 @@
  uri: "quay.io/go-skynet/local-ai-backends:master-gpu-rocm-hipblas-neutts"
  mirrors:
    - localai/localai-backends:master-gpu-rocm-hipblas-neutts
+- !!merge <<: *neutts
+  name: "nvidia-l4t-arm64-neutts-development"
+  uri: "quay.io/go-skynet/local-ai-backends:master-nvidia-l4t-arm64-neutts"
+  mirrors:
+    - localai/localai-backends:master-nvidia-l4t-arm64-neutts
 - !!merge <<: *mlx
  name: "mlx-development"
  uri: "quay.io/go-skynet/local-ai-backends:master-metal-darwin-arm64-mlx"
@@ -1635,169 +1605,3 @@
  uri: "quay.io/go-skynet/local-ai-backends:master-nvidia-l4t-cuda-13-arm64-vibevoice"
  mirrors:
    - localai/localai-backends:master-nvidia-l4t-cuda-13-arm64-vibevoice
-## qwen-tts
- !!merge <<: *qwen-tts
-  name: "qwen-tts-development"
-  capabilities:
-    nvidia: "cuda12-qwen-tts-development"
-    intel: "intel-qwen-tts-development"
-    amd: "rocm-qwen-tts-development"
-    nvidia-l4t: "nvidia-l4t-qwen-tts-development"
-    default: "cpu-qwen-tts-development"
-    nvidia-cuda-13: "cuda13-qwen-tts-development"
-    nvidia-cuda-12: "cuda12-qwen-tts-development"
-    nvidia-l4t-cuda-12: "nvidia-l4t-qwen-tts-development"
-    nvidia-l4t-cuda-13: "cuda13-nvidia-l4t-arm64-qwen-tts-development"
- !!merge <<: *qwen-tts
-  name: "cpu-qwen-tts"
-  uri: "quay.io/go-skynet/local-ai-backends:latest-cpu-qwen-tts"
-  mirrors:
-    - localai/localai-backends:latest-cpu-qwen-tts
- !!merge <<: *qwen-tts
-  name: "cpu-qwen-tts-development"
-  uri: "quay.io/go-skynet/local-ai-backends:master-cpu-qwen-tts"
-  mirrors:
-    - localai/localai-backends:master-cpu-qwen-tts
- !!merge <<: *qwen-tts
-  name: "cuda12-qwen-tts"
-  uri: "quay.io/go-skynet/local-ai-backends:latest-gpu-nvidia-cuda-12-qwen-tts"
-  mirrors:
-    - localai/localai-backends:latest-gpu-nvidia-cuda-12-qwen-tts
- !!merge <<: *qwen-tts
-  name: "cuda12-qwen-tts-development"
-  uri: "quay.io/go-skynet/local-ai-backends:master-gpu-nvidia-cuda-12-qwen-tts"
-  mirrors:
-    - localai/localai-backends:master-gpu-nvidia-cuda-12-qwen-tts
- !!merge <<: *qwen-tts
-  name: "cuda13-qwen-tts"
-  uri: "quay.io/go-skynet/local-ai-backends:latest-gpu-nvidia-cuda-13-qwen-tts"
-  mirrors:
-    - localai/localai-backends:latest-gpu-nvidia-cuda-13-qwen-tts
- !!merge <<: *qwen-tts
-  name: "cuda13-qwen-tts-development"
-  uri: "quay.io/go-skynet/local-ai-backends:master-gpu-nvidia-cuda-13-qwen-tts"
-  mirrors:
-    - localai/localai-backends:master-gpu-nvidia-cuda-13-qwen-tts
- !!merge <<: *qwen-tts
-  name: "intel-qwen-tts"
-  uri: "quay.io/go-skynet/local-ai-backends:latest-gpu-intel-qwen-tts"
-  mirrors:
-    - localai/localai-backends:latest-gpu-intel-qwen-tts
- !!merge <<: *qwen-tts
-  name: "intel-qwen-tts-development"
-  uri: "quay.io/go-skynet/local-ai-backends:master-gpu-intel-qwen-tts"
-  mirrors:
-    - localai/localai-backends:master-gpu-intel-qwen-tts
- !!merge <<: *qwen-tts
-  name: "rocm-qwen-tts"
-  uri: "quay.io/go-skynet/local-ai-backends:latest-gpu-rocm-hipblas-qwen-tts"
-  mirrors:
-    - localai/localai-backends:latest-gpu-rocm-hipblas-qwen-tts
- !!merge <<: *qwen-tts
-  name: "rocm-qwen-tts-development"
-  uri: "quay.io/go-skynet/local-ai-backends:master-gpu-rocm-hipblas-qwen-tts"
-  mirrors:
-    - localai/localai-backends:master-gpu-rocm-hipblas-qwen-tts
- !!merge <<: *qwen-tts
-  name: "nvidia-l4t-qwen-tts"
-  uri: "quay.io/go-skynet/local-ai-backends:latest-nvidia-l4t-qwen-tts"
-  mirrors:
-    - localai/localai-backends:latest-nvidia-l4t-qwen-tts
- !!merge <<: *qwen-tts
-  name: "nvidia-l4t-qwen-tts-development"
-  uri: "quay.io/go-skynet/local-ai-backends:master-nvidia-l4t-qwen-tts"
-  mirrors:
-    - localai/localai-backends:master-nvidia-l4t-qwen-tts
- !!merge <<: *qwen-tts
-  name: "cuda13-nvidia-l4t-arm64-qwen-tts"
-  uri: "quay.io/go-skynet/local-ai-backends:latest-nvidia-l4t-cuda-13-arm64-qwen-tts"
-  mirrors:
-    - localai/localai-backends:latest-nvidia-l4t-cuda-13-arm64-qwen-tts
- !!merge <<: *qwen-tts
-  name: "cuda13-nvidia-l4t-arm64-qwen-tts-development"
-  uri: "quay.io/go-skynet/local-ai-backends:master-nvidia-l4t-cuda-13-arm64-qwen-tts"
-  mirrors:
-    - localai/localai-backends:master-nvidia-l4t-cuda-13-arm64-qwen-tts
-## pocket-tts
- !!merge <<: *pocket-tts
-  name: "pocket-tts-development"
-  capabilities:
-    nvidia: "cuda12-pocket-tts-development"
-    intel: "intel-pocket-tts-development"
-    amd: "rocm-pocket-tts-development"
-    nvidia-l4t: "nvidia-l4t-pocket-tts-development"
-    default: "cpu-pocket-tts-development"
-    nvidia-cuda-13: "cuda13-pocket-tts-development"
-    nvidia-cuda-12: "cuda12-pocket-tts-development"
-    nvidia-l4t-cuda-12: "nvidia-l4t-pocket-tts-development"
-    nvidia-l4t-cuda-13: "cuda13-nvidia-l4t-arm64-pocket-tts-development"
- !!merge <<: *pocket-tts
-  name: "cpu-pocket-tts"
-  uri: "quay.io/go-skynet/local-ai-backends:latest-cpu-pocket-tts"
-  mirrors:
-    - localai/localai-backends:latest-cpu-pocket-tts
- !!merge <<: *pocket-tts
-  name: "cpu-pocket-tts-development"
-  uri: "quay.io/go-skynet/local-ai-backends:master-cpu-pocket-tts"
-  mirrors:
-    - localai/localai-backends:master-cpu-pocket-tts
- !!merge <<: *pocket-tts
-  name: "cuda12-pocket-tts"
-  uri: "quay.io/go-skynet/local-ai-backends:latest-gpu-nvidia-cuda-12-pocket-tts"
-  mirrors:
-    - localai/localai-backends:latest-gpu-nvidia-cuda-12-pocket-tts
- !!merge <<: *pocket-tts
-  name: "cuda12-pocket-tts-development"
-  uri: "quay.io/go-skynet/local-ai-backends:master-gpu-nvidia-cuda-12-pocket-tts"
-  mirrors:
-    - localai/localai-backends:master-gpu-nvidia-cuda-12-pocket-tts
- !!merge <<: *pocket-tts
-  name: "cuda13-pocket-tts"
-  uri: "quay.io/go-skynet/local-ai-backends:latest-gpu-nvidia-cuda-13-pocket-tts"
-  mirrors:
-    - localai/localai-backends:latest-gpu-nvidia-cuda-13-pocket-tts
- !!merge <<: *pocket-tts
-  name: "cuda13-pocket-tts-development"
-  uri: "quay.io/go-skynet/local-ai-backends:master-gpu-nvidia-cuda-13-pocket-tts"
-  mirrors:
-    - localai/localai-backends:master-gpu-nvidia-cuda-13-pocket-tts
- !!merge <<: *pocket-tts
-  name: "intel-pocket-tts"
-  uri: "quay.io/go-skynet/local-ai-backends:latest-gpu-intel-pocket-tts"
-  mirrors:
-    - localai/localai-backends:latest-gpu-intel-pocket-tts
- !!merge <<: *pocket-tts
-  name: "intel-pocket-tts-development"
-  uri: "quay.io/go-skynet/local-ai-backends:master-gpu-intel-pocket-tts"
-  mirrors:
-    - localai/localai-backends:master-gpu-intel-pocket-tts
- !!merge <<: *pocket-tts
-  name: "rocm-pocket-tts"
-  uri: "quay.io/go-skynet/local-ai-backends:latest-gpu-rocm-hipblas-pocket-tts"
-  mirrors:
-    - localai/localai-backends:latest-gpu-rocm-hipblas-pocket-tts
- !!merge <<: *pocket-tts
-  name: "rocm-pocket-tts-development"
-  uri: "quay.io/go-skynet/local-ai-backends:master-gpu-rocm-hipblas-pocket-tts"
-  mirrors:
-    - localai/localai-backends:master-gpu-rocm-hipblas-pocket-tts
- !!merge <<: *pocket-tts
-  name: "nvidia-l4t-pocket-tts"
-  uri: "quay.io/go-skynet/local-ai-backends:latest-nvidia-l4t-pocket-tts"
-  mirrors:
-    - localai/localai-backends:latest-nvidia-l4t-pocket-tts
- !!merge <<: *pocket-tts
-  name: "nvidia-l4t-pocket-tts-development"
-  uri: "quay.io/go-skynet/local-ai-backends:master-nvidia-l4t-pocket-tts"
-  mirrors:
-    - localai/localai-backends:master-nvidia-l4t-pocket-tts
- !!merge <<: *pocket-tts
-  name: "cuda13-nvidia-l4t-arm64-pocket-tts"
-  uri: "quay.io/go-skynet/local-ai-backends:latest-nvidia-l4t-cuda-13-arm64-pocket-tts"
-  mirrors:
-    - localai/localai-backends:latest-nvidia-l4t-cuda-13-arm64-pocket-tts
- !!merge <<: *pocket-tts
-  name: "cuda13-nvidia-l4t-arm64-pocket-tts-development"
-  uri: "quay.io/go-skynet/local-ai-backends:master-nvidia-l4t-cuda-13-arm64-pocket-tts"
-  mirrors:
-    - localai/localai-backends:master-nvidia-l4t-cuda-13-arm64-pocket-tts
--- a/backend/python/bark/requirements-intel.txt
+++ b/backend/python/bark/requirements-intel.txt
@@ -1,6 +1,8 @@
--extra-index-url https://download.pytorch.org/whl/xpu
-torch
-torchaudio
+--extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/
+intel-extension-for-pytorch==2.8.10+xpu
+torch==2.3.1+cxx11.abi
+torchaudio==2.3.1+cxx11.abi
+oneccl_bind_pt==2.3.100+xpu
 optimum[openvino]
 setuptools
 transformers
--- a/backend/python/chatterbox/requirements-intel.txt
+++ b/backend/python/chatterbox/requirements-intel.txt
@@ -1,6 +1,7 @@
--extra-index-url https://download.pytorch.org/whl/xpu
-torch
-torchaudio
+--extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/
+intel-extension-for-pytorch==2.3.110+xpu
+torch==2.3.1+cxx11.abi
+torchaudio==2.3.1+cxx11.abi
 transformers
 numpy>=1.24.0,<1.26.0
 # https://github.com/mudler/LocalAI/pull/6240#issuecomment-3329518289
--- a/backend/python/common/libbackend.sh
+++ b/backend/python/common/libbackend.sh
@@ -398,7 +398,7 @@ function runProtogen() {
 # NOTE: for BUILD_PROFILE==intel, this function does NOT automatically use the Intel python package index.
 # you may want to add the following line to a requirements-intel.txt if you use one:
 #
-# --index-url https://download.pytorch.org/whl/xpu
+# --index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/
 #
 # If you need to add extra flags into the pip install command you can do so by setting the variable EXTRA_PIP_INSTALL_FLAGS
 # before calling installRequirements.  For example:
--- a/backend/python/common/template/requirements-intel.txt
+++ b/backend/python/common/template/requirements-intel.txt
@@ -1,4 +1,5 @@
--extra-index-url https://download.pytorch.org/whl/xpu
+--extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/
+intel-extension-for-pytorch==2.8.10+xpu
 torch==2.8.0
 oneccl_bind_pt==2.8.0+xpu
 optimum[openvino]
--- a/backend/python/coqui/requirements-intel.txt
+++ b/backend/python/coqui/requirements-intel.txt
@@ -1,6 +1,8 @@
--extra-index-url https://download.pytorch.org/whl/xpu
-torch==2.8.0+xpu
-torchaudio==2.8.0+xpu
+--extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/
+intel-extension-for-pytorch==2.3.110+xpu
+torch==2.3.1+cxx11.abi
+torchaudio==2.3.1+cxx11.abi
+oneccl_bind_pt==2.3.100+xpu
 optimum[openvino]
 setuptools
 transformers==4.48.3
--- a/backend/python/diffusers/backend.py
+++ b/backend/python/diffusers/backend.py
@@ -41,10 +41,6 @@ from optimum.quanto import freeze, qfloat8, quantize
 from transformers import T5EncoderModel
 from safetensors.torch import load_file

-# Import LTX-2 specific utilities
-from diffusers.pipelines.ltx2.export_utils import encode_video as ltx2_encode_video
-from diffusers import LTX2VideoTransformer3DModel, GGUFQuantizationConfig
-
 _ONE_DAY_IN_SECONDS = 60 * 60 * 24
 COMPEL = os.environ.get("COMPEL", "0") == "1"
 XPU = os.environ.get("XPU", "0") == "1"
@@ -294,104 +290,6 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
                pipe.enable_model_cpu_offload()
            return pipe

-        # LTX2ImageToVideoPipeline - needs img2vid flag, CPU offload, and special handling
-        if pipeline_type == "LTX2ImageToVideoPipeline":
-            self.img2vid = True
-            self.ltx2_pipeline = True
-            
-            # Check if loading from single file (GGUF)
-            if fromSingleFile and LTX2VideoTransformer3DModel is not None:
-                _, single_file_ext = os.path.splitext(modelFile)
-                if single_file_ext == ".gguf":
-                    # Load transformer from single GGUF file with quantization
-                    transformer_kwargs = {}
-                    quantization_config = GGUFQuantizationConfig(compute_dtype=torchType)
-                    transformer_kwargs["quantization_config"] = quantization_config
-                    
-                    transformer = LTX2VideoTransformer3DModel.from_single_file(
-                        modelFile,
-                        config=request.Model,  # Use request.Model as the config/model_id
-                        subfolder="transformer",
-                        **transformer_kwargs,
-                    )
-                    
-                    # Load pipeline with custom transformer
-                    pipe = load_diffusers_pipeline(
-                        class_name="LTX2ImageToVideoPipeline",
-                        model_id=request.Model,
-                        transformer=transformer,
-                        torch_dtype=torchType,
-                    )
-                else:
-                    # Single file but not GGUF - use standard single file loading
-                    pipe = load_diffusers_pipeline(
-                        class_name="LTX2ImageToVideoPipeline",
-                        model_id=modelFile,
-                        from_single_file=True,
-                        torch_dtype=torchType,
-                    )
-            else:
-                # Standard loading from pretrained
-                pipe = load_diffusers_pipeline(
-                    class_name="LTX2ImageToVideoPipeline",
-                    model_id=request.Model,
-                    torch_dtype=torchType,
-                    variant=variant
-                )
-            
-            if not DISABLE_CPU_OFFLOAD:
-                pipe.enable_model_cpu_offload()
-            return pipe
-
-        # LTX2Pipeline - text-to-video pipeline, needs txt2vid flag, CPU offload, and special handling
-        if pipeline_type == "LTX2Pipeline":
-            self.txt2vid = True
-            self.ltx2_pipeline = True
-            
-            # Check if loading from single file (GGUF)
-            if fromSingleFile and LTX2VideoTransformer3DModel is not None:
-                _, single_file_ext = os.path.splitext(modelFile)
-                if single_file_ext == ".gguf":
-                    # Load transformer from single GGUF file with quantization
-                    transformer_kwargs = {}
-                    quantization_config = GGUFQuantizationConfig(compute_dtype=torchType)
-                    transformer_kwargs["quantization_config"] = quantization_config
-                    
-                    transformer = LTX2VideoTransformer3DModel.from_single_file(
-                        modelFile,
-                        config=request.Model,  # Use request.Model as the config/model_id
-                        subfolder="transformer",
-                        **transformer_kwargs,
-                    )
-                    
-                    # Load pipeline with custom transformer
-                    pipe = load_diffusers_pipeline(
-                        class_name="LTX2Pipeline",
-                        model_id=request.Model,
-                        transformer=transformer,
-                        torch_dtype=torchType,
-                    )
-                else:
-                    # Single file but not GGUF - use standard single file loading
-                    pipe = load_diffusers_pipeline(
-                        class_name="LTX2Pipeline",
-                        model_id=modelFile,
-                        from_single_file=True,
-                        torch_dtype=torchType,
-                    )
-            else:
-                # Standard loading from pretrained
-                pipe = load_diffusers_pipeline(
-                    class_name="LTX2Pipeline",
-                    model_id=request.Model,
-                    torch_dtype=torchType,
-                    variant=variant
-                )
-            
-            if not DISABLE_CPU_OFFLOAD:
-                pipe.enable_model_cpu_offload()
-            return pipe
-
        # ================================================================
        # Dynamic pipeline loading - the default path for most pipelines
        # Uses the dynamic loader to instantiate any pipeline by class name
@@ -506,9 +404,6 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
            fromSingleFile = request.Model.startswith("http") or request.Model.startswith("/") or local
            self.img2vid = False
            self.txt2vid = False
-            self.ltx2_pipeline = False
-
-            print(f"LoadModel: PipelineType from request: {request.PipelineType}", file=sys.stderr)

            # Load pipeline using dynamic loader
            # Special cases that require custom initialization are handled first
@@ -519,8 +414,6 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
                torchType=torchType,
                variant=variant
            )
-            
-            print(f"LoadModel: After loading - ltx2_pipeline: {self.ltx2_pipeline}, img2vid: {self.img2vid}, txt2vid: {self.txt2vid}, PipelineType: {self.PipelineType}", file=sys.stderr)

            if CLIPSKIP and request.CLIPSkip != 0:
                self.clip_skip = request.CLIPSkip
@@ -758,20 +651,14 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
        try:
            prompt = request.prompt
            if not prompt:
-                print(f"GenerateVideo: No prompt provided for video generation.", file=sys.stderr)
                return backend_pb2.Result(success=False, message="No prompt provided for video generation")

-            # Debug: Print raw request values
-            print(f"GenerateVideo: Raw request values - num_frames: {request.num_frames}, fps: {request.fps}, cfg_scale: {request.cfg_scale}, step: {request.step}", file=sys.stderr)
-
            # Set default values from request or use defaults
            num_frames = request.num_frames if request.num_frames > 0 else 81
            fps = request.fps if request.fps > 0 else 16
            cfg_scale = request.cfg_scale if request.cfg_scale > 0 else 4.0
            num_inference_steps = request.step if request.step > 0 else 40
            
-            print(f"GenerateVideo: Using values - num_frames: {num_frames}, fps: {fps}, cfg_scale: {cfg_scale}, num_inference_steps: {num_inference_steps}", file=sys.stderr)
-            
            # Prepare generation parameters
            kwargs = {
                "prompt": prompt,
@@ -797,86 +684,9 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
                kwargs["end_image"] = load_image(request.end_image)

            print(f"Generating video with {kwargs=}", file=sys.stderr)
-            print(f"GenerateVideo: Pipeline type: {self.PipelineType}, ltx2_pipeline flag: {self.ltx2_pipeline}", file=sys.stderr)

            # Generate video frames based on pipeline type
-            if self.ltx2_pipeline or self.PipelineType in ["LTX2Pipeline", "LTX2ImageToVideoPipeline"]:
-                # LTX-2 generation with audio (supports both text-to-video and image-to-video)
-                # Determine if this is text-to-video (no image) or image-to-video (has image)
-                has_image = bool(request.start_image)
-                
-                # Remove image-related parameters that might have been added earlier
-                kwargs.pop("start_image", None)
-                kwargs.pop("end_image", None)
-                
-                # LTX2ImageToVideoPipeline uses 'image' parameter for image-to-video
-                # LTX2Pipeline (text-to-video) doesn't need an image parameter
-                if has_image:
-                    # Image-to-video: use 'image' parameter
-                    if self.PipelineType == "LTX2ImageToVideoPipeline":
-                        image = load_image(request.start_image)
-                        kwargs["image"] = image
-                        print(f"LTX-2: Using image-to-video mode with image", file=sys.stderr)
-                    else:
-                        # If pipeline type is LTX2Pipeline but we have an image, we can't do image-to-video
-                        return backend_pb2.Result(success=False, message="LTX2Pipeline does not support image-to-video. Use LTX2ImageToVideoPipeline for image-to-video generation.")
-                else:
-                    # Text-to-video: no image parameter needed
-                    # Ensure no image-related kwargs are present
-                    kwargs.pop("image", None)
-                    print(f"LTX-2: Using text-to-video mode (no image)", file=sys.stderr)
-                
-                # LTX-2 uses 'frame_rate' instead of 'fps'
-                frame_rate = float(fps)
-                kwargs["frame_rate"] = frame_rate
-                
-                # LTX-2 requires output_type="np" and return_dict=False
-                kwargs["output_type"] = "np"
-                kwargs["return_dict"] = False
-                
-                # Generate video and audio
-                print(f"LTX-2: Generating with kwargs: {kwargs}", file=sys.stderr)
-                try:
-                    video, audio = self.pipe(**kwargs)
-                    print(f"LTX-2: Generated video shape: {video.shape}, audio shape: {audio.shape}", file=sys.stderr)
-                except Exception as e:
-                    print(f"LTX-2: Error during pipe() call: {e}", file=sys.stderr)
-                    traceback.print_exc()
-                    return backend_pb2.Result(success=False, message=f"Error generating video with LTX-2 pipeline: {e}")
-                
-                # Convert video to uint8 format
-                video = (video * 255).round().astype("uint8")
-                video = torch.from_numpy(video)
-                
-                print(f"LTX-2: Converting video, shape after conversion: {video.shape}", file=sys.stderr)
-                print(f"LTX-2: Audio sample rate: {self.pipe.vocoder.config.output_sampling_rate}", file=sys.stderr)
-                print(f"LTX-2: Output path: {request.dst}", file=sys.stderr)
-                
-                # Use LTX-2's encode_video function which handles audio
-                try:
-                    ltx2_encode_video(
-                        video[0],
-                        fps=frame_rate,
-                        audio=audio[0].float().cpu(),
-                        audio_sample_rate=self.pipe.vocoder.config.output_sampling_rate,
-                        output_path=request.dst,
-                    )
-                    # Verify file was created and has content
-                    import os
-                    if os.path.exists(request.dst):
-                        file_size = os.path.getsize(request.dst)
-                        print(f"LTX-2: Video file created successfully, size: {file_size} bytes", file=sys.stderr)
-                        if file_size == 0:
-                            return backend_pb2.Result(success=False, message=f"Video file was created but is empty (0 bytes). Check LTX-2 encode_video function.")
-                    else:
-                        return backend_pb2.Result(success=False, message=f"Video file was not created at {request.dst}")
-                except Exception as e:
-                    print(f"LTX-2: Error encoding video: {e}", file=sys.stderr)
-                    traceback.print_exc()
-                    return backend_pb2.Result(success=False, message=f"Error encoding video: {e}")
-                
-                return backend_pb2.Result(message="Video generated successfully", success=True)
-            elif self.PipelineType == "WanPipeline":
+            if self.PipelineType == "WanPipeline":
                # WAN2.2 text-to-video generation
                output = self.pipe(**kwargs)
                frames = output.frames[0]  # WAN2.2 returns frames in this format
@@ -915,23 +725,11 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
                output = self.pipe(**kwargs)
                frames = output.frames[0]
            else:
-                print(f"GenerateVideo: Pipeline {self.PipelineType} does not match any known video pipeline handler", file=sys.stderr)
                return backend_pb2.Result(success=False, message=f"Pipeline {self.PipelineType} does not support video generation")

-            # Export video (for non-LTX-2 pipelines)
-            print(f"GenerateVideo: Exporting video to {request.dst} with fps={fps}", file=sys.stderr)
+            # Export video
            export_to_video(frames, request.dst, fps=fps)
            
-            # Verify file was created
-            import os
-            if os.path.exists(request.dst):
-                file_size = os.path.getsize(request.dst)
-                print(f"GenerateVideo: Video file created, size: {file_size} bytes", file=sys.stderr)
-                if file_size == 0:
-                    return backend_pb2.Result(success=False, message=f"Video file was created but is empty (0 bytes)")
-            else:
-                return backend_pb2.Result(success=False, message=f"Video file was not created at {request.dst}")
-            
            return backend_pb2.Result(message="Video generated successfully", success=True)

        except Exception as err:
--- a/backend/python/diffusers/requirements-intel.txt
+++ b/backend/python/diffusers/requirements-intel.txt
@@ -1,6 +1,8 @@
--extra-index-url https://download.pytorch.org/whl/xpu
-torch
-torchvision
+--extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/
+intel-extension-for-pytorch==2.3.110+xpu
+torch==2.5.1+cxx11.abi
+torchvision==0.20.1+cxx11.abi
+oneccl_bind_pt==2.8.0+xpu
 optimum[openvino]
 setuptools
 git+https://github.com/huggingface/diffusers
--- a/backend/python/diffusers/requirements.txt
+++ b/backend/python/diffusers/requirements.txt
@@ -3,4 +3,3 @@ grpcio==1.76.0
 pillow
 protobuf
 certifi
-av
--- a/backend/python/faster-whisper/requirements-intel.txt
+++ b/backend/python/faster-whisper/requirements-intel.txt
@@ -1,4 +1,6 @@
--extra-index-url https://download.pytorch.org/whl/xpu
-torch
+--extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/
+intel-extension-for-pytorch==2.3.110+xpu
+torch==2.3.1+cxx11.abi
+oneccl_bind_pt==2.3.100+xpu
 optimum[openvino]
 faster-whisper
--- a/backend/python/kokoro/requirements-intel.txt
+++ b/backend/python/kokoro/requirements-intel.txt
@@ -1,6 +1,8 @@
--extra-index-url https://download.pytorch.org/whl/xpu
-torch
-torchaudio
+--extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/
+intel-extension-for-pytorch==2.8.10+xpu
+torch==2.5.1+cxx11.abi
+oneccl_bind_pt==2.8.0+xpu
+torchaudio==2.5.1+cxx11.abi
 optimum[openvino]
 setuptools
 transformers==4.48.3
--- a/backend/python/pocket-tts/Makefile
+++ b/backend/python/pocket-tts/Makefile
@@ -1,23 +0,0 @@
-.PHONY: pocket-tts
-pocket-tts:
-	bash install.sh
-
-.PHONY: run
-run: pocket-tts
-	@echo "Running pocket-tts..."
-	bash run.sh
-	@echo "pocket-tts run."
-
-.PHONY: test
-test: pocket-tts
-	@echo "Testing pocket-tts..."
-	bash test.sh
-	@echo "pocket-tts tested."
-
-.PHONY: protogen-clean
-protogen-clean:
-	$(RM) backend_pb2_grpc.py backend_pb2.py
-
-.PHONY: clean
-clean: protogen-clean
-	rm -rf venv __pycache__
--- a/backend/python/pocket-tts/backend.py
+++ b/backend/python/pocket-tts/backend.py
@@ -1,255 +0,0 @@
-#!/usr/bin/env python3
-"""
-This is an extra gRPC server of LocalAI for Pocket TTS
-"""
-from concurrent import futures
-import time
-import argparse
-import signal
-import sys
-import os
-import traceback
-import scipy.io.wavfile
-import backend_pb2
-import backend_pb2_grpc
-import torch
-from pocket_tts import TTSModel
-
-import grpc
-
-def is_float(s):
-    """Check if a string can be converted to float."""
-    try:
-        float(s)
-        return True
-    except ValueError:
-        return False
-
-def is_int(s):
-    """Check if a string can be converted to int."""
-    try:
-        int(s)
-        return True
-    except ValueError:
-        return False
-
-_ONE_DAY_IN_SECONDS = 60 * 60 * 24
-
-# If MAX_WORKERS are specified in the environment use it, otherwise default to 1
-MAX_WORKERS = int(os.environ.get('PYTHON_GRPC_MAX_WORKERS', '1'))
-
-# Implement the BackendServicer class with the service methods
-class BackendServicer(backend_pb2_grpc.BackendServicer):
-    """
-    BackendServicer is the class that implements the gRPC service
-    """
-    def Health(self, request, context):
-        return backend_pb2.Reply(message=bytes("OK", 'utf-8'))
-    
-    def LoadModel(self, request, context):
-        # Get device
-        if torch.cuda.is_available():
-            print("CUDA is available", file=sys.stderr)
-            device = "cuda"
-        else:
-            print("CUDA is not available", file=sys.stderr)
-            device = "cpu"
-        mps_available = hasattr(torch.backends, "mps") and torch.backends.mps.is_available()
-        if mps_available:
-            device = "mps"
-        if not torch.cuda.is_available() and request.CUDA:
-            return backend_pb2.Result(success=False, message="CUDA is not available")
-
-        # Normalize potential 'mpx' typo to 'mps'
-        if device == "mpx":
-            print("Note: device 'mpx' detected, treating it as 'mps'.", file=sys.stderr)
-            device = "mps"
-        
-        # Validate mps availability if requested
-        if device == "mps" and not torch.backends.mps.is_available():
-            print("Warning: MPS not available. Falling back to CPU.", file=sys.stderr)
-            device = "cpu"
-
-        self.device = device
-
-        options = request.Options
-
-        # empty dict
-        self.options = {}
-
-        # The options are a list of strings in this form optname:optvalue
-        # We are storing all the options in a dict so we can use it later when
-        # generating the audio
-        for opt in options:
-            if ":" not in opt:
-                continue
-            key, value = opt.split(":", 1)  # Split only on first colon
-            # if value is a number, convert it to the appropriate type
-            if is_float(value):
-                value = float(value)
-            elif is_int(value):
-                value = int(value)
-            elif value.lower() in ["true", "false"]:
-                value = value.lower() == "true"
-            self.options[key] = value
-
-        # Default voice for caching
-        self.default_voice_url = self.options.get("default_voice", None)
-        self._voice_cache = {}
-
-        try:
-            print("Loading Pocket TTS model", file=sys.stderr)
-            self.tts_model = TTSModel.load_model()
-            print(f"Model loaded successfully. Sample rate: {self.tts_model.sample_rate}", file=sys.stderr)
-
-            # Pre-load default voice if specified
-            if self.default_voice_url:
-                try:
-                    print(f"Pre-loading default voice: {self.default_voice_url}", file=sys.stderr)
-                    voice_state = self.tts_model.get_state_for_audio_prompt(self.default_voice_url)
-                    self._voice_cache[self.default_voice_url] = voice_state
-                    print("Default voice loaded successfully", file=sys.stderr)
-                except Exception as e:
-                    print(f"Warning: Failed to pre-load default voice: {e}", file=sys.stderr)
-
-        except Exception as err:
-            return backend_pb2.Result(success=False, message=f"Unexpected {err=}, {type(err)=}")
-        
-        return backend_pb2.Result(message="Model loaded successfully", success=True)
-
-    def _get_voice_state(self, voice_input):
-        """
-        Get voice state from cache or load it.
-        voice_input can be:
-        - HuggingFace URL (e.g., hf://kyutai/tts-voices/alba-mackenna/casual.wav)
-        - Local file path
-        - None (use default)
-        """
-        # Use default if no voice specified
-        if not voice_input:
-            voice_input = self.default_voice_url
-
-        if not voice_input:
-            return None
-
-        # Check cache first
-        if voice_input in self._voice_cache:
-            return self._voice_cache[voice_input]
-
-        # Load voice state
-        try:
-            print(f"Loading voice from: {voice_input}", file=sys.stderr)
-            voice_state = self.tts_model.get_state_for_audio_prompt(voice_input)
-            self._voice_cache[voice_input] = voice_state
-            return voice_state
-        except Exception as e:
-            print(f"Error loading voice from {voice_input}: {e}", file=sys.stderr)
-            return None
-
-    def TTS(self, request, context):
-        try:
-            # Determine voice input
-            # Priority: request.voice > AudioPath (from ModelOptions) > default
-            voice_input = None
-            
-            if request.voice:
-                voice_input = request.voice
-            elif hasattr(request, 'AudioPath') and request.AudioPath:
-                # Use AudioPath as voice file
-                if os.path.isabs(request.AudioPath):
-                    voice_input = request.AudioPath
-                elif hasattr(request, 'ModelFile') and request.ModelFile:
-                    model_file_base = os.path.dirname(request.ModelFile)
-                    voice_input = os.path.join(model_file_base, request.AudioPath)
-                elif hasattr(request, 'ModelPath') and request.ModelPath:
-                    voice_input = os.path.join(request.ModelPath, request.AudioPath)
-                else:
-                    voice_input = request.AudioPath
-
-            # Get voice state
-            voice_state = self._get_voice_state(voice_input)
-            if voice_state is None:
-                return backend_pb2.Result(
-                    success=False,
-                    message=f"Voice not found or failed to load: {voice_input}. Please provide a valid voice URL or file path."
-                )
-
-            # Prepare text
-            text = request.text.strip()
-
-            if not text:
-                return backend_pb2.Result(
-                    success=False,
-                    message="Text is empty"
-                )
-
-            print(f"Generating audio for text: {text[:50]}...", file=sys.stderr)
-
-            # Generate audio
-            audio = self.tts_model.generate_audio(voice_state, text)
-
-            # Audio is a 1D torch tensor containing PCM data
-            if audio is None or audio.numel() == 0:
-                return backend_pb2.Result(
-                    success=False,
-                    message="No audio generated"
-                )
-
-            # Save audio to file
-            output_path = request.dst
-            if not output_path:
-                output_path = "/tmp/pocket-tts-output.wav"
-
-            # Ensure output directory exists
-            output_dir = os.path.dirname(output_path)
-            if output_dir and not os.path.exists(output_dir):
-                os.makedirs(output_dir, exist_ok=True)
-
-            # Convert torch tensor to numpy and save
-            audio_numpy = audio.numpy()
-            scipy.io.wavfile.write(output_path, self.tts_model.sample_rate, audio_numpy)
-            print(f"Saved audio to {output_path}", file=sys.stderr)
-
-        except Exception as err:
-            print(f"Error in TTS: {err}", file=sys.stderr)
-            print(traceback.format_exc(), file=sys.stderr)
-            return backend_pb2.Result(success=False, message=f"Unexpected {err=}, {type(err)=}")
-        
-        return backend_pb2.Result(success=True)
-
-def serve(address):
-    server = grpc.server(futures.ThreadPoolExecutor(max_workers=MAX_WORKERS),
-        options=[
-            ('grpc.max_message_length', 50 * 1024 * 1024),  # 50MB
-            ('grpc.max_send_message_length', 50 * 1024 * 1024),  # 50MB
-            ('grpc.max_receive_message_length', 50 * 1024 * 1024),  # 50MB
-        ])
-    backend_pb2_grpc.add_BackendServicer_to_server(BackendServicer(), server)
-    server.add_insecure_port(address)
-    server.start()
-    print("Server started. Listening on: " + address, file=sys.stderr)
-
-    # Define the signal handler function
-    def signal_handler(sig, frame):
-        print("Received termination signal. Shutting down...")
-        server.stop(0)
-        sys.exit(0)
-
-    # Set the signal handlers for SIGINT and SIGTERM
-    signal.signal(signal.SIGINT, signal_handler)
-    signal.signal(signal.SIGTERM, signal_handler)
-
-    try:
-        while True:
-            time.sleep(_ONE_DAY_IN_SECONDS)
-    except KeyboardInterrupt:
-        server.stop(0)
-
-if __name__ == "__main__":
-    parser = argparse.ArgumentParser(description="Run the gRPC server.")
-    parser.add_argument(
-        "--addr", default="localhost:50051", help="The address to bind the server to."
-    )
-    args = parser.parse_args()
-
-    serve(args.addr)
--- a/backend/python/pocket-tts/install.sh
+++ b/backend/python/pocket-tts/install.sh
@@ -1,30 +0,0 @@
-#!/bin/bash
-set -e
-
-backend_dir=$(dirname $0)
-if [ -d $backend_dir/common ]; then
-    source $backend_dir/common/libbackend.sh
-else
-    source $backend_dir/../common/libbackend.sh
-fi
-
-# This is here because the Intel pip index is broken and returns 200 status codes for every package name, it just doesn't return any package links.
-# This makes uv think that the package exists in the Intel pip index, and by default it stops looking at other pip indexes once it finds a match.
-# We need uv to continue falling through to the pypi default index to find optimum[openvino] in the pypi index
-# the --upgrade actually allows us to *downgrade* torch to the version provided in the Intel pip index
-if [ "x${BUILD_PROFILE}" == "xintel" ]; then
-    EXTRA_PIP_INSTALL_FLAGS+=" --upgrade --index-strategy=unsafe-first-match"
-fi
-
-# Use python 3.12 for l4t
-if [ "x${BUILD_PROFILE}" == "xl4t13" ]; then
-  PYTHON_VERSION="3.12"
-  PYTHON_PATCH="12"
-  PY_STANDALONE_TAG="20251120"
-fi
-
-if [ "x${BUILD_PROFILE}" == "xl4t12" ]; then
-    USE_PIP=true
-fi
-
-installRequirements
--- a/backend/python/pocket-tts/protogen.sh
+++ b/backend/python/pocket-tts/protogen.sh
@@ -1,11 +0,0 @@
-#!/bin/bash
-set -e
-
-backend_dir=$(dirname $0)
-if [ -d $backend_dir/common ]; then
-    source $backend_dir/common/libbackend.sh
-else
-    source $backend_dir/../common/libbackend.sh
-fi
-
-python3 -m grpc_tools.protoc -I../.. -I./ --python_out=. --grpc_python_out=. backend.proto
--- a/backend/python/pocket-tts/requirements-cpu.txt
+++ b/backend/python/pocket-tts/requirements-cpu.txt
@@ -1,4 +0,0 @@
--extra-index-url https://download.pytorch.org/whl/cpu
-pocket-tts
-scipy
-torch
--- a/backend/python/pocket-tts/requirements-cublas12.txt
+++ b/backend/python/pocket-tts/requirements-cublas12.txt
@@ -1,4 +0,0 @@
--extra-index-url https://download.pytorch.org/whl/cu121
-pocket-tts
-scipy
-torch
--- a/backend/python/pocket-tts/requirements-cublas13.txt
+++ b/backend/python/pocket-tts/requirements-cublas13.txt
@@ -1,4 +0,0 @@
--extra-index-url https://download.pytorch.org/whl/cu130
-pocket-tts
-scipy
-torch
--- a/backend/python/pocket-tts/requirements-hipblas.txt
+++ b/backend/python/pocket-tts/requirements-hipblas.txt
@@ -1,4 +0,0 @@
--extra-index-url https://download.pytorch.org/whl/rocm6.3
-pocket-tts
-scipy
-torch==2.7.1+rocm6.3
--- a/backend/python/pocket-tts/requirements-intel.txt
+++ b/backend/python/pocket-tts/requirements-intel.txt
@@ -1,4 +0,0 @@
--extra-index-url https://download.pytorch.org/whl/xpu
-pocket-tts
-scipy
-torch
--- a/backend/python/pocket-tts/requirements-l4t12.txt
+++ b/backend/python/pocket-tts/requirements-l4t12.txt
@@ -1,4 +0,0 @@
--extra-index-url https://pypi.jetson-ai-lab.io/jp6/cu129/
-pocket-tts
-scipy
-torch
--- a/backend/python/pocket-tts/requirements-l4t13.txt
+++ b/backend/python/pocket-tts/requirements-l4t13.txt
@@ -1,4 +0,0 @@
--extra-index-url https://download.pytorch.org/whl/cu130
-pocket-tts
-scipy
-torch
--- a/backend/python/pocket-tts/requirements-mps.txt
+++ b/backend/python/pocket-tts/requirements-mps.txt
@@ -1,4 +0,0 @@
-pocket-tts
-scipy
-torch==2.7.1
-torchvision==0.22.1
--- a/backend/python/pocket-tts/requirements.txt
+++ b/backend/python/pocket-tts/requirements.txt
@@ -1,4 +0,0 @@
-grpcio==1.71.0
-protobuf
-certifi
-packaging==24.1
--- a/backend/python/pocket-tts/run.sh
+++ b/backend/python/pocket-tts/run.sh
@@ -1,9 +0,0 @@
-#!/bin/bash
-backend_dir=$(dirname $0)
-if [ -d $backend_dir/common ]; then
-    source $backend_dir/common/libbackend.sh
-else
-    source $backend_dir/../common/libbackend.sh
-fi
-
-startBackend $@
--- a/backend/python/pocket-tts/test.py
+++ b/backend/python/pocket-tts/test.py
@@ -1,141 +0,0 @@
-"""
-A test script to test the gRPC service
-"""
-import unittest
-import subprocess
-import time
-import os
-import tempfile
-import backend_pb2
-import backend_pb2_grpc
-
-import grpc
-
-
-class TestBackendServicer(unittest.TestCase):
-    """
-    TestBackendServicer is the class that tests the gRPC service
-    """
-    def setUp(self):
-        """
-        This method sets up the gRPC service by starting the server
-        """
-        self.service = subprocess.Popen(["python3", "backend.py", "--addr", "localhost:50051"])
-        time.sleep(30)
-
-    def tearDown(self) -> None:
-        """
-        This method tears down the gRPC service by terminating the server
-        """
-        self.service.terminate()
-        self.service.wait()
-
-    def test_server_startup(self):
-        """
-        This method tests if the server starts up successfully
-        """
-        try:
-            self.setUp()
-            with grpc.insecure_channel("localhost:50051") as channel:
-                stub = backend_pb2_grpc.BackendStub(channel)
-                response = stub.Health(backend_pb2.HealthMessage())
-                self.assertEqual(response.message, b'OK')
-        except Exception as err:
-            print(err)
-            self.fail("Server failed to start")
-        finally:
-            self.tearDown()
-
-    def test_load_model(self):
-        """
-        This method tests if the model is loaded successfully
-        """
-        try:
-            self.setUp()
-            with grpc.insecure_channel("localhost:50051") as channel:
-                stub = backend_pb2_grpc.BackendStub(channel)
-                response = stub.LoadModel(backend_pb2.ModelOptions())
-                print(response)
-                self.assertTrue(response.success)
-                self.assertEqual(response.message, "Model loaded successfully")
-        except Exception as err:
-            print(err)
-            self.fail("LoadModel service failed")
-        finally:
-            self.tearDown()
-
-    def test_tts_with_hf_voice(self):
-        """
-        This method tests TTS generation with HuggingFace voice URL
-        """
-        try:
-            self.setUp()
-            with grpc.insecure_channel("localhost:50051") as channel:
-                stub = backend_pb2_grpc.BackendStub(channel)
-                # Load model
-                response = stub.LoadModel(backend_pb2.ModelOptions())
-                self.assertTrue(response.success)
-                
-                # Create temporary output file
-                with tempfile.NamedTemporaryFile(suffix='.wav', delete=False) as tmp_file:
-                    output_path = tmp_file.name
-                
-                # Test TTS with HuggingFace voice URL
-                tts_request = backend_pb2.TTSRequest(
-                    text="Hello world, this is a test.",
-                    dst=output_path,
-                    voice="azelma"
-                )
-                tts_response = stub.TTS(tts_request)
-                self.assertTrue(tts_response.success)
-                
-                # Verify output file exists and is not empty
-                self.assertTrue(os.path.exists(output_path))
-                self.assertGreater(os.path.getsize(output_path), 0)
-                
-                # Cleanup
-                os.unlink(output_path)
-        except Exception as err:
-            print(err)
-            self.fail("TTS service failed")
-        finally:
-            self.tearDown()
-
-    def test_tts_with_default_voice(self):
-        """
-        This method tests TTS generation with default voice (via AudioPath in LoadModel)
-        """
-        try:
-            self.setUp()
-            with grpc.insecure_channel("localhost:50051") as channel:
-                stub = backend_pb2_grpc.BackendStub(channel)
-                # Load model with default voice
-                load_request = backend_pb2.ModelOptions(
-                    Options=["default_voice:azelma"]
-                )
-                response = stub.LoadModel(load_request)
-                self.assertTrue(response.success)
-                
-                # Create temporary output file
-                with tempfile.NamedTemporaryFile(suffix='.wav', delete=False) as tmp_file:
-                    output_path = tmp_file.name
-                
-                # Test TTS without specifying voice (should use default)
-                tts_request = backend_pb2.TTSRequest(
-                    text="Hello world, this is a test.",
-                    dst=output_path
-                )
-                tts_response = stub.TTS(tts_request)
-                self.assertTrue(tts_response.success)
-                
-                # Verify output file exists and is not empty
-                self.assertTrue(os.path.exists(output_path))
-                self.assertGreater(os.path.getsize(output_path), 0)
-                
-                # Cleanup
-                os.unlink(output_path)
-        except Exception as err:
-            print(err)
-            self.fail("TTS service with default voice failed")
-        finally:
-            self.tearDown()
--- a/backend/python/pocket-tts/test.sh
+++ b/backend/python/pocket-tts/test.sh
@@ -1,11 +0,0 @@
-#!/bin/bash
-set -e
-
-backend_dir=$(dirname $0)
-if [ -d $backend_dir/common ]; then
-    source $backend_dir/common/libbackend.sh
-else
-    source $backend_dir/../common/libbackend.sh
-fi
-
-runUnittests
--- a/backend/python/qwen-tts/Makefile
+++ b/backend/python/qwen-tts/Makefile
@@ -1,23 +0,0 @@
-.PHONY: qwen-tts
-qwen-tts:
-	bash install.sh
-
-.PHONY: run
-run: qwen-tts
-	@echo "Running qwen-tts..."
-	bash run.sh
-	@echo "qwen-tts run."
-
-.PHONY: test
-test: qwen-tts
-	@echo "Testing qwen-tts..."
-	bash test.sh
-	@echo "qwen-tts tested."
-
-.PHONY: protogen-clean
-protogen-clean:
-	$(RM) backend_pb2_grpc.py backend_pb2.py
-
-.PHONY: clean
-clean: protogen-clean
-	rm -rf venv __pycache__
--- a/backend/python/qwen-tts/backend.py
+++ b/backend/python/qwen-tts/backend.py
@@ -1,475 +0,0 @@
-#!/usr/bin/env python3
-"""
-This is an extra gRPC server of LocalAI for Qwen3-TTS
-"""
-from concurrent import futures
-import time
-import argparse
-import signal
-import sys
-import os
-import copy
-import traceback
-from pathlib import Path
-import backend_pb2
-import backend_pb2_grpc
-import torch
-import soundfile as sf
-from qwen_tts import Qwen3TTSModel
-
-import grpc
-
-def is_float(s):
-    """Check if a string can be converted to float."""
-    try:
-        float(s)
-        return True
-    except ValueError:
-        return False
-
-def is_int(s):
-    """Check if a string can be converted to int."""
-    try:
-        int(s)
-        return True
-    except ValueError:
-        return False
-
-_ONE_DAY_IN_SECONDS = 60 * 60 * 24
-
-# If MAX_WORKERS are specified in the environment use it, otherwise default to 1
-MAX_WORKERS = int(os.environ.get('PYTHON_GRPC_MAX_WORKERS', '1'))
-
-
-# Implement the BackendServicer class with the service methods
-class BackendServicer(backend_pb2_grpc.BackendServicer):
-    """
-    BackendServicer is the class that implements the gRPC service
-    """
-    def Health(self, request, context):
-        return backend_pb2.Reply(message=bytes("OK", 'utf-8'))
-    
-    def LoadModel(self, request, context):
-        # Get device
-        if torch.cuda.is_available():
-            print("CUDA is available", file=sys.stderr)
-            device = "cuda"
-        else:
-            print("CUDA is not available", file=sys.stderr)
-            device = "cpu"
-        mps_available = hasattr(torch.backends, "mps") and torch.backends.mps.is_available()
-        if mps_available:
-            device = "mps"
-        if not torch.cuda.is_available() and request.CUDA:
-            return backend_pb2.Result(success=False, message="CUDA is not available")
-
-        # Normalize potential 'mpx' typo to 'mps'
-        if device == "mpx":
-            print("Note: device 'mpx' detected, treating it as 'mps'.", file=sys.stderr)
-            device = "mps"
-        
-        # Validate mps availability if requested
-        if device == "mps" and not torch.backends.mps.is_available():
-            print("Warning: MPS not available. Falling back to CPU.", file=sys.stderr)
-            device = "cpu"
-
-        self.device = device
-        self._torch_device = torch.device(device)
-
-        options = request.Options
-
-        # empty dict
-        self.options = {}
-
-        # The options are a list of strings in this form optname:optvalue
-        # We are storing all the options in a dict so we can use it later when
-        # generating the audio
-        for opt in options:
-            if ":" not in opt:
-                continue
-            key, value = opt.split(":", 1)  # Split only on first colon
-            # if value is a number, convert it to the appropriate type
-            if is_float(value):
-                value = float(value)
-            elif is_int(value):
-                value = int(value)
-            elif value.lower() in ["true", "false"]:
-                value = value.lower() == "true"
-            self.options[key] = value
-
-        # Get model path from request
-        model_path = request.Model
-        if not model_path:
-            model_path = "Qwen/Qwen3-TTS-12Hz-1.7B-CustomVoice"
-
-        # Determine model type from model path or options
-        self.model_type = self.options.get("model_type", None)
-        if not self.model_type:
-            if "CustomVoice" in model_path:
-                self.model_type = "CustomVoice"
-            elif "VoiceDesign" in model_path:
-                self.model_type = "VoiceDesign"
-            elif "Base" in model_path or "0.6B" in model_path or "1.7B" in model_path:
-                self.model_type = "Base"  # VoiceClone model
-            else:
-                # Default to CustomVoice
-                self.model_type = "CustomVoice"
-
-        # Cache for voice clone prompts
-        self._voice_clone_cache = {}
-
-        # Store AudioPath, ModelFile, and ModelPath from LoadModel request
-        # These are used later in TTS for VoiceClone mode
-        self.audio_path = request.AudioPath if hasattr(request, 'AudioPath') and request.AudioPath else None
-        self.model_file = request.ModelFile if hasattr(request, 'ModelFile') and request.ModelFile else None
-        self.model_path = request.ModelPath if hasattr(request, 'ModelPath') and request.ModelPath else None
-
-        # Decide dtype & attention implementation
-        if self.device == "mps":
-            load_dtype = torch.float32  # MPS requires float32
-            device_map = None
-            attn_impl_primary = "sdpa"  # flash_attention_2 not supported on MPS
-        elif self.device == "cuda":
-            load_dtype = torch.bfloat16
-            device_map = "cuda"
-            attn_impl_primary = "flash_attention_2"
-        else:  # cpu
-            load_dtype = torch.float32
-            device_map = "cpu"
-            attn_impl_primary = "sdpa"
-
-        print(f"Using device: {self.device}, torch_dtype: {load_dtype}, attn_implementation: {attn_impl_primary}, model_type: {self.model_type}", file=sys.stderr)
-        print(f"Loading model from: {model_path}", file=sys.stderr)
-
-        # Load model with device-specific logic
-        # Common parameters for all devices
-        load_kwargs = {
-            "dtype": load_dtype,
-            "attn_implementation": attn_impl_primary,
-            "trust_remote_code": True,  # Required for qwen-tts models
-        }
-        
-        try:
-            if self.device == "mps":
-                load_kwargs["device_map"] = None  # load then move
-                self.model = Qwen3TTSModel.from_pretrained(model_path, **load_kwargs)
-                self.model.to("mps")
-            elif self.device == "cuda":
-                load_kwargs["device_map"] = device_map
-                self.model = Qwen3TTSModel.from_pretrained(model_path, **load_kwargs)
-            else:  # cpu
-                load_kwargs["device_map"] = device_map
-                self.model = Qwen3TTSModel.from_pretrained(model_path, **load_kwargs)
-        except Exception as e:
-            error_msg = str(e)
-            print(f"[ERROR] Loading model: {type(e).__name__}: {error_msg}", file=sys.stderr)
-            print(traceback.format_exc(), file=sys.stderr)
-            
-            # Check if it's a missing feature extractor/tokenizer error
-            if "speech_tokenizer" in error_msg or "preprocessor_config.json" in error_msg or "feature extractor" in error_msg.lower():
-                print("\n[ERROR] Model files appear to be incomplete. This usually means:", file=sys.stderr)
-                print("  1. The model download was interrupted or incomplete", file=sys.stderr)
-                print("  2. The model cache is corrupted", file=sys.stderr)
-                print("\nTo fix this, try:", file=sys.stderr)
-                print(f"  rm -rf ~/.cache/huggingface/hub/models--Qwen--Qwen3-TTS-*", file=sys.stderr)
-                print("  Then re-run to trigger a fresh download.", file=sys.stderr)
-                print("\nAlternatively, try using a different model variant:", file=sys.stderr)
-                print("  - Qwen/Qwen3-TTS-12Hz-1.7B-CustomVoice", file=sys.stderr)
-                print("  - Qwen/Qwen3-TTS-12Hz-1.7B-VoiceDesign", file=sys.stderr)
-                print("  - Qwen/Qwen3-TTS-12Hz-1.7B-Base", file=sys.stderr)
-            
-            if attn_impl_primary == 'flash_attention_2':
-                print("\nTrying to use SDPA instead of flash_attention_2...", file=sys.stderr)
-                load_kwargs["attn_implementation"] = 'sdpa'
-                try:
-                    if self.device == "mps":
-                        load_kwargs["device_map"] = None
-                        self.model = Qwen3TTSModel.from_pretrained(model_path, **load_kwargs)
-                        self.model.to("mps")
-                    else:
-                        load_kwargs["device_map"] = (self.device if self.device in ("cuda", "cpu") else None)
-                        self.model = Qwen3TTSModel.from_pretrained(model_path, **load_kwargs)
-                except Exception as e2:
-                    print(f"[ERROR] Failed to load with SDPA: {type(e2).__name__}: {e2}", file=sys.stderr)
-                    print(traceback.format_exc(), file=sys.stderr)
-                    raise e2
-            else:
-                raise e
-
-        print(f"Model loaded successfully: {model_path}", file=sys.stderr)
-
-        return backend_pb2.Result(message="Model loaded successfully", success=True)
-
-    def _detect_mode(self, request):
-        """Detect which mode to use based on request parameters."""
-        # Priority: VoiceClone > VoiceDesign > CustomVoice
-
-        # model_type explicitly set
-        if self.model_type == "CustomVoice":
-            return "CustomVoice"
-        if self.model_type == "VoiceClone":
-            return "VoiceClone"
-        if self.model_type == "VoiceDesign":
-            return "VoiceDesign"
-
-        # VoiceClone: AudioPath is provided (from LoadModel, stored in self.audio_path)
-        if self.audio_path:
-            return "VoiceClone"
-        
-        # VoiceDesign: instruct option is provided
-        if "instruct" in self.options and self.options["instruct"]:
-            return "VoiceDesign"
-        
-        # Default to CustomVoice
-        return "CustomVoice"
-
-    def _get_ref_audio_path(self, request):
-        """Get reference audio path from stored AudioPath (from LoadModel)."""
-        if not self.audio_path:
-            return None
-        
-        # If absolute path, use as-is
-        if os.path.isabs(self.audio_path):
-            return self.audio_path
-        
-        # Try relative to ModelFile
-        if self.model_file:
-            model_file_base = os.path.dirname(self.model_file)
-            ref_path = os.path.join(model_file_base, self.audio_path)
-            if os.path.exists(ref_path):
-                return ref_path
-        
-        # Try relative to ModelPath
-        if self.model_path:
-            ref_path = os.path.join(self.model_path, self.audio_path)
-            if os.path.exists(ref_path):
-                return ref_path
-        
-        # Return as-is (might be URL or base64)
-        return self.audio_path
-
-    def _get_voice_clone_prompt(self, request, ref_audio, ref_text):
-        """Get or create voice clone prompt, with caching."""
-        cache_key = f"{ref_audio}:{ref_text}"
-        
-        if cache_key not in self._voice_clone_cache:
-            print(f"Creating voice clone prompt from {ref_audio}", file=sys.stderr)
-            try:
-                prompt_items = self.model.create_voice_clone_prompt(
-                    ref_audio=ref_audio,
-                    ref_text=ref_text,
-                    x_vector_only_mode=self.options.get("x_vector_only_mode", False),
-                )
-                self._voice_clone_cache[cache_key] = prompt_items
-            except Exception as e:
-                print(f"Error creating voice clone prompt: {e}", file=sys.stderr)
-                print(traceback.format_exc(), file=sys.stderr)
-                return None
-        
-        return self._voice_clone_cache[cache_key]
-
-    def TTS(self, request, context):
-        try:
-            # Check if dst is provided
-            if not request.dst:
-                return backend_pb2.Result(
-                    success=False,
-                    message="dst (output path) is required"
-                )
-            
-            # Prepare text
-            text = request.text.strip()
-            if not text:
-                return backend_pb2.Result(
-                    success=False,
-                    message="Text is empty"
-                )
-
-            # Get language (auto-detect if not provided)
-            language = request.language if hasattr(request, 'language') and request.language else None
-            if not language or language == "":
-                language = "Auto"  # Auto-detect language
-
-            # Detect mode
-            mode = self._detect_mode(request)
-            print(f"Detected mode: {mode}", file=sys.stderr)
-
-            # Get generation parameters from options
-            max_new_tokens = self.options.get("max_new_tokens", None)
-            top_p = self.options.get("top_p", None)
-            temperature = self.options.get("temperature", None)
-            do_sample = self.options.get("do_sample", None)
-
-            # Prepare generation kwargs
-            generation_kwargs = {}
-            if max_new_tokens is not None:
-                generation_kwargs["max_new_tokens"] = max_new_tokens
-            if top_p is not None:
-                generation_kwargs["top_p"] = top_p
-            if temperature is not None:
-                generation_kwargs["temperature"] = temperature
-            if do_sample is not None:
-                generation_kwargs["do_sample"] = do_sample
-
-            instruct = self.options.get("instruct", "")
-            if instruct is not None and instruct != "":
-                generation_kwargs["instruct"] = instruct
-
-            # Generate audio based on mode
-            if mode == "VoiceClone":
-                # VoiceClone mode
-                ref_audio = self._get_ref_audio_path(request)
-                if not ref_audio:
-                    return backend_pb2.Result(
-                        success=False,
-                        message="AudioPath is required for VoiceClone mode"
-                    )
-                
-                ref_text = self.options.get("ref_text", None)
-                if not ref_text:
-                    # Try to get from request if available
-                    if hasattr(request, 'ref_text') and request.ref_text:
-                        ref_text = request.ref_text
-                    else:
-                        # x_vector_only_mode doesn't require ref_text
-                        if not self.options.get("x_vector_only_mode", False):
-                            return backend_pb2.Result(
-                                success=False,
-                                message="ref_text is required for VoiceClone mode (or set x_vector_only_mode=true)"
-                            )
-
-                # Check if we should use cached prompt
-                use_cached_prompt = self.options.get("use_cached_prompt", True)
-                voice_clone_prompt = None
-                
-                if use_cached_prompt:
-                    voice_clone_prompt = self._get_voice_clone_prompt(request, ref_audio, ref_text)
-                    if voice_clone_prompt is None:
-                        return backend_pb2.Result(
-                            success=False,
-                            message="Failed to create voice clone prompt"
-                        )
-
-                if voice_clone_prompt:
-                    # Use cached prompt
-                    wavs, sr = self.model.generate_voice_clone(
-                        text=text,
-                        language=language,
-                        voice_clone_prompt=voice_clone_prompt,
-                        **generation_kwargs
-                    )
-                else:
-                    # Create prompt on-the-fly
-                    wavs, sr = self.model.generate_voice_clone(
-                        text=text,
-                        language=language,
-                        ref_audio=ref_audio,
-                        ref_text=ref_text,
-                        x_vector_only_mode=self.options.get("x_vector_only_mode", False),
-                        **generation_kwargs
-                    )
-
-            elif mode == "VoiceDesign":
-                # VoiceDesign mode
-                if not instruct:
-                    return backend_pb2.Result(
-                        success=False,
-                        message="instruct option is required for VoiceDesign mode"
-                    )
-
-                wavs, sr = self.model.generate_voice_design(
-                    text=text,
-                    language=language,
-                    instruct=instruct,
-                    **generation_kwargs
-                )
-
-            else:
-                # CustomVoice mode (default)
-                speaker = request.voice if request.voice else None
-                if not speaker:
-                    # Try to get from options
-                    speaker = self.options.get("speaker", None)
-                    if not speaker:
-                        # Use default speaker
-                        speaker = "Vivian"
-                        print(f"No speaker specified, using default: {speaker}", file=sys.stderr)
-
-                # Validate speaker if model supports it
-                if hasattr(self.model, 'get_supported_speakers'):
-                    try:
-                        supported_speakers = self.model.get_supported_speakers()
-                        if speaker not in supported_speakers:
-                            print(f"Warning: Speaker '{speaker}' not in supported list. Available: {supported_speakers}", file=sys.stderr)
-                            # Try to find a close match (case-insensitive)
-                            speaker_lower = speaker.lower()
-                            for sup_speaker in supported_speakers:
-                                if sup_speaker.lower() == speaker_lower:
-                                    speaker = sup_speaker
-                                    print(f"Using matched speaker: {speaker}", file=sys.stderr)
-                                    break
-                    except Exception as e:
-                        print(f"Warning: Could not get supported speakers: {e}", file=sys.stderr)
-
-                wavs, sr = self.model.generate_custom_voice(
-                    text=text,
-                    language=language,
-                    speaker=speaker,
-                    **generation_kwargs
-                )
-
-            # Save output
-            if wavs is not None and len(wavs) > 0:
-                # wavs is a list, take first element
-                audio_data = wavs[0] if isinstance(wavs, list) else wavs
-                sf.write(request.dst, audio_data, sr)
-                print(f"Saved output to {request.dst}", file=sys.stderr)
-            else:
-                return backend_pb2.Result(
-                    success=False,
-                    message="No audio output generated"
-                )
-
-        except Exception as err:
-            print(f"Error in TTS: {err}", file=sys.stderr)
-            print(traceback.format_exc(), file=sys.stderr)
-            return backend_pb2.Result(success=False, message=f"Unexpected {err=}, {type(err)=}")
-        
-        return backend_pb2.Result(success=True)
-
-def serve(address):
-    server = grpc.server(futures.ThreadPoolExecutor(max_workers=MAX_WORKERS),
-        options=[
-            ('grpc.max_message_length', 50 * 1024 * 1024),  # 50MB
-            ('grpc.max_send_message_length', 50 * 1024 * 1024),  # 50MB
-            ('grpc.max_receive_message_length', 50 * 1024 * 1024),  # 50MB
-        ])
-    backend_pb2_grpc.add_BackendServicer_to_server(BackendServicer(), server)
-    server.add_insecure_port(address)
-    server.start()
-    print("Server started. Listening on: " + address, file=sys.stderr)
-
-    # Define the signal handler function
-    def signal_handler(sig, frame):
-        print("Received termination signal. Shutting down...")
-        server.stop(0)
-        sys.exit(0)
-
-    # Set the signal handlers for SIGINT and SIGTERM
-    signal.signal(signal.SIGINT, signal_handler)
-    signal.signal(signal.SIGTERM, signal_handler)
-
-    try:
-        while True:
-            time.sleep(_ONE_DAY_IN_SECONDS)
-    except KeyboardInterrupt:
-        server.stop(0)
-
-if __name__ == "__main__":
-    parser = argparse.ArgumentParser(description="Run the gRPC server.")
-    parser.add_argument(
-        "--addr", default="localhost:50051", help="The address to bind the server to."
-    )
-    args = parser.parse_args()
-
-    serve(args.addr)
--- a/backend/python/qwen-tts/install.sh
+++ b/backend/python/qwen-tts/install.sh
@@ -1,13 +0,0 @@
-#!/bin/bash
-set -e
-
-EXTRA_PIP_INSTALL_FLAGS="--no-build-isolation"
-
-backend_dir=$(dirname $0)
-if [ -d $backend_dir/common ]; then
-    source $backend_dir/common/libbackend.sh
-else
-    source $backend_dir/../common/libbackend.sh
-fi
-
-installRequirements
--- a/backend/python/qwen-tts/requirements-cpu.txt
+++ b/backend/python/qwen-tts/requirements-cpu.txt
@@ -1,5 +0,0 @@
--extra-index-url https://download.pytorch.org/whl/cpu
-torch
-torchaudio
-qwen-tts
-sox
--- a/backend/python/qwen-tts/requirements-cublas12-after.txt
+++ b/backend/python/qwen-tts/requirements-cublas12-after.txt
@@ -1 +0,0 @@
-flash-attn
--- a/backend/python/qwen-tts/requirements-cublas12.txt
+++ b/backend/python/qwen-tts/requirements-cublas12.txt
@@ -1,5 +0,0 @@
--extra-index-url https://download.pytorch.org/whl/cu121
-torch
-torchaudio
-qwen-tts
-sox
--- a/backend/python/qwen-tts/requirements-cublas13.txt
+++ b/backend/python/qwen-tts/requirements-cublas13.txt
@@ -1,5 +0,0 @@
--extra-index-url https://download.pytorch.org/whl/cu130
-torch
-torchaudio
-qwen-tts
-sox
--- a/backend/python/qwen-tts/requirements-hipblas.txt
+++ b/backend/python/qwen-tts/requirements-hipblas.txt
@@ -1,5 +0,0 @@
--extra-index-url https://download.pytorch.org/whl/rocm6.3
-torch==2.7.1+rocm6.3
-torchaudio==2.7.1+rocm6.3
-qwen-tts
-sox
--- a/backend/python/qwen-tts/requirements-intel-after.txt
+++ b/backend/python/qwen-tts/requirements-intel-after.txt
@@ -1 +0,0 @@
-flash-attn
--- a/backend/python/qwen-tts/requirements-intel.txt
+++ b/backend/python/qwen-tts/requirements-intel.txt
@@ -1,5 +0,0 @@
--extra-index-url https://download.pytorch.org/whl/xpu
-torch
-torchaudio
-qwen-tts
-sox
--- a/backend/python/qwen-tts/requirements-l4t12.txt
+++ b/backend/python/qwen-tts/requirements-l4t12.txt
@@ -1,5 +0,0 @@
--extra-index-url https://pypi.jetson-ai-lab.io/jp6/cu129/
-torch
-torchaudio
-qwen-tts
-sox
--- a/backend/python/qwen-tts/requirements-l4t13.txt
+++ b/backend/python/qwen-tts/requirements-l4t13.txt
@@ -1,5 +0,0 @@
--extra-index-url https://download.pytorch.org/whl/cu130
-torch
-torchaudio
-qwen-tts
-sox
--- a/backend/python/qwen-tts/requirements-mps.txt
+++ b/backend/python/qwen-tts/requirements-mps.txt
@@ -1,4 +0,0 @@
-torch==2.7.1
-torchaudio==0.22.1
-qwen-tts
-sox
--- a/backend/python/qwen-tts/requirements.txt
+++ b/backend/python/qwen-tts/requirements.txt
@@ -1,6 +0,0 @@
-grpcio==1.71.0
-protobuf
-certifi
-packaging==24.1
-soundfile
-setuptools
--- a/backend/python/qwen-tts/run.sh
+++ b/backend/python/qwen-tts/run.sh
@@ -1,9 +0,0 @@
-#!/bin/bash
-backend_dir=$(dirname $0)
-if [ -d $backend_dir/common ]; then
-    source $backend_dir/common/libbackend.sh
-else
-    source $backend_dir/../common/libbackend.sh
-fi
-
-startBackend $@
--- a/backend/python/qwen-tts/test.py
+++ b/backend/python/qwen-tts/test.py
@@ -1,98 +0,0 @@
-"""
-A test script to test the gRPC service
-"""
-import unittest
-import subprocess
-import time
-import os
-import sys
-import tempfile
-import threading
-import backend_pb2
-import backend_pb2_grpc
-
-import grpc
-
-
-class TestBackendServicer(unittest.TestCase):
-    """
-    TestBackendServicer is the class that tests the gRPC service
-    """
-    def setUp(self):
-        """
-        This method sets up the gRPC service by starting the server
-        """
-        self.service = subprocess.Popen(
-            ["python3", "backend.py", "--addr", "localhost:50051"],
-            stdout=subprocess.PIPE,
-            stderr=subprocess.PIPE,
-            text=True
-        )
-        time.sleep(5)
-
-    def tearDown(self) -> None:
-        """
-        This method tears down the gRPC service by terminating the server
-        """
-        self.service.terminate()
-        try:
-            stdout, stderr = self.service.communicate(timeout=5)
-            # Output should already be printed by threads, but print any remaining
-            if stdout:
-                print("=== REMAINING STDOUT ===")
-                print(stdout)
-            if stderr:
-                print("=== REMAINING STDERR ===")
-                print(stderr)
-        except subprocess.TimeoutExpired:
-            self.service.kill()
-            stdout, stderr = self.service.communicate()
-            if stdout:
-                print("=== REMAINING STDOUT ===")
-                print(stdout)
-            if stderr:
-                print("=== REMAINING STDERR ===")
-                print(stderr)
-
-    def test_tts(self):
-        """
-        This method tests if the TTS generation works successfully
-        """
-        try:
-            self.setUp()
-            with grpc.insecure_channel("localhost:50051") as channel:
-                stub = backend_pb2_grpc.BackendStub(channel)
-                # Allow up to 10 minutes for model download on first run
-                response = stub.LoadModel(
-                    backend_pb2.ModelOptions(Model="Qwen/Qwen3-TTS-12Hz-0.6B-CustomVoice"),
-                    timeout=600.0
-                )
-                self.assertTrue(response.success)
-                
-                # Create temporary output file
-                with tempfile.NamedTemporaryFile(suffix='.wav', delete=False) as tmp_file:
-                    output_path = tmp_file.name
-                
-                tts_request = backend_pb2.TTSRequest(
-                    text="Hello, this is a test of the qwen-tts backend.",
-                    voice="Vivian",
-                    dst=output_path
-                )
-                # Allow up to 2 minutes for TTS generation
-                tts_response = stub.TTS(tts_request, timeout=120.0)
-                self.assertIsNotNone(tts_response)
-                self.assertTrue(tts_response.success)
-                
-                # Verify output file exists and is not empty
-                self.assertTrue(os.path.exists(output_path))
-                self.assertGreater(os.path.getsize(output_path), 0)
-                
-                # Cleanup
-                os.unlink(output_path)
-        except Exception as err:
-            print(f"Exception: {err}", file=sys.stderr)
-            # Give threads a moment to flush any remaining output
-            time.sleep(1)
-            self.fail("TTS service failed")
-        finally:
-            self.tearDown()
--- a/backend/python/qwen-tts/test.sh
+++ b/backend/python/qwen-tts/test.sh
@@ -1,11 +0,0 @@
-#!/bin/bash
-set -e
-
-backend_dir=$(dirname $0)
-if [ -d $backend_dir/common ]; then
-    source $backend_dir/common/libbackend.sh
-else
-    source $backend_dir/../common/libbackend.sh
-fi
-
-runUnittests
--- a/backend/python/rerankers/requirements-intel.txt
+++ b/backend/python/rerankers/requirements-intel.txt
@@ -1,7 +1,9 @@
--extra-index-url https://download.pytorch.org/whl/xpu
+--extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/
+intel-extension-for-pytorch==2.3.110+xpu
 transformers
 accelerate
-torch
+torch==2.3.1+cxx11.abi
+oneccl_bind_pt==2.8.0+xpu
 rerankers[transformers]
 optimum[openvino]
 setuptools
--- a/backend/python/rfdetr/requirements-intel.txt
+++ b/backend/python/rfdetr/requirements-intel.txt
@@ -1,6 +1,8 @@
--extra-index-url https://download.pytorch.org/whl/xpu
-torch
-torchvision
+--extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/
+intel-extension-for-pytorch==2.3.110+xpu
+torch==2.3.1+cxx11.abi
+torchvision==0.18.1+cxx11.abi
+oneccl_bind_pt==2.3.100+xpu
 optimum[openvino]
 setuptools
 rfdetr
--- a/backend/python/transformers/requirements-cpu.txt
+++ b/backend/python/transformers/requirements-cpu.txt
@@ -6,4 +6,4 @@ transformers
 bitsandbytes
 outetts
 sentence-transformers==5.2.0
-protobuf==6.33.4
+protobuf==6.33.2
--- a/backend/python/transformers/requirements-cublas12.txt
+++ b/backend/python/transformers/requirements-cublas12.txt
@@ -6,4 +6,4 @@ transformers
 bitsandbytes
 outetts
 sentence-transformers==5.2.0
-protobuf==6.33.4
+protobuf==6.33.2
--- a/backend/python/transformers/requirements-cublas13.txt
+++ b/backend/python/transformers/requirements-cublas13.txt
@@ -6,4 +6,4 @@ transformers
 bitsandbytes
 outetts
 sentence-transformers==5.2.0
-protobuf==6.33.4
+protobuf==6.33.2
--- a/backend/python/transformers/requirements-hipblas.txt
+++ b/backend/python/transformers/requirements-hipblas.txt
@@ -8,4 +8,4 @@ bitsandbytes
 outetts
 bitsandbytes
 sentence-transformers==5.2.0
-protobuf==6.33.4
+protobuf==6.33.2
--- a/backend/python/transformers/requirements-intel.txt
+++ b/backend/python/transformers/requirements-intel.txt
@@ -1,10 +1,13 @@
--extra-index-url https://download.pytorch.org/whl/xpu
-torch
+--extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/
+intel-extension-for-pytorch==2.3.110+xpu
+torch==2.5.1+cxx11.abi
+oneccl_bind_pt==2.8.0+xpu
 optimum[openvino]
 llvmlite==0.43.0
 numba==0.60.0
 transformers
+intel-extension-for-transformers
 bitsandbytes
 outetts
 sentence-transformers==5.2.0
-protobuf==6.33.4
+protobuf==6.33.2
--- a/backend/python/transformers/requirements.txt
+++ b/backend/python/transformers/requirements.txt
@@ -1,5 +1,5 @@
 grpcio==1.76.0
-protobuf==6.33.4
+protobuf==6.33.2
 certifi
 setuptools
 scipy==1.15.1
--- a/backend/python/vibevoice/requirements-intel.txt
+++ b/backend/python/vibevoice/requirements-intel.txt
@@ -1,6 +1,8 @@
--extra-index-url https://download.pytorch.org/whl/xpu
-torch
-torchvision
+--extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/
+intel-extension-for-pytorch==2.3.110+xpu
+torch==2.5.1+cxx11.abi
+torchvision==0.20.1+cxx11.abi
+oneccl_bind_pt==2.8.0+xpu
 optimum[openvino]
 setuptools
 git+https://github.com/huggingface/diffusers
--- a/backend/python/vllm/requirements-intel.txt
+++ b/backend/python/vllm/requirements-intel.txt
@@ -1,7 +1,10 @@
 --extra-index-url https://download.pytorch.org/whl/xpu
+--extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/
+intel-extension-for-pytorch==2.7.10+xpu
 accelerate
-torch
+torch==2.7.0+xpu
 transformers
 optimum[openvino]
 setuptools
-bitsandbytes
+bitsandbytes
+oneccl_bind_pt==2.7.0+xpu
--- a/core/backend/llm.go
+++ b/core/backend/llm.go
@@ -61,18 +61,6 @@ func ModelInference(ctx context.Context, s string, messages schema.Messages, ima
 		return nil, err
 	}

-	// Detect thinking support after model load (only if not already detected)
-	// This needs to happen after LoadModel succeeds so the backend can render templates
-	if (c.ReasoningConfig.DisableReasoning == nil && c.ReasoningConfig.DisableReasoningTagPrefill == nil) && c.TemplateConfig.UseTokenizerTemplate {
-		modelOpts := grpcModelOpts(*c, o.SystemState.Model.ModelsPath)
-		config.DetectThinkingSupportFromBackend(ctx, c, inferenceModel, modelOpts)
-		// Update the config in the loader so it persists for future requests
-		cl.UpdateModelConfig(c.Name, func(cfg *config.ModelConfig) {
-			cfg.ReasoningConfig.DisableReasoning = c.ReasoningConfig.DisableReasoning
-			cfg.ReasoningConfig.DisableReasoningTagPrefill = c.ReasoningConfig.DisableReasoningTagPrefill
-		})
-	}
-
 	var protoMessages []*proto.Message
 	// if we are using the tokenizer template, we need to convert the messages to proto messages
 	// unless the prompt has already been tokenized (non-chat endpoints + functions)
--- a/core/cli/run.go
+++ b/core/cli/run.go
@@ -56,7 +56,7 @@ type RunCMD struct {
 	UseSubtleKeyComparison             bool     `env:"LOCALAI_SUBTLE_KEY_COMPARISON" default:"false" help:"If true, API Key validation comparisons will be performed using constant-time comparisons rather than simple equality. This trades off performance on each request for resiliancy against timing attacks." group:"hardening"`
 	DisableApiKeyRequirementForHttpGet bool     `env:"LOCALAI_DISABLE_API_KEY_REQUIREMENT_FOR_HTTP_GET" default:"false" help:"If true, a valid API key is not required to issue GET requests to portions of the web ui. This should only be enabled in secure testing environments" group:"hardening"`
 	DisableMetricsEndpoint             bool     `env:"LOCALAI_DISABLE_METRICS_ENDPOINT,DISABLE_METRICS_ENDPOINT" default:"false" help:"Disable the /metrics endpoint" group:"api"`
-	HttpGetExemptedEndpoints           []string `env:"LOCALAI_HTTP_GET_EXEMPTED_ENDPOINTS" default:"^/$,^/browse/?$,^/talk/?$,^/p2p/?$,^/chat/?$,^/image/?$,^/text2image/?$,^/tts/?$,^/static/.*$,^/swagger.*$" help:"If LOCALAI_DISABLE_API_KEY_REQUIREMENT_FOR_HTTP_GET is overriden to true, this is the list of endpoints to exempt. Only adjust this in case of a security incident or as a result of a personal security posture review" group:"hardening"`
+	HttpGetExemptedEndpoints           []string `env:"LOCALAI_HTTP_GET_EXEMPTED_ENDPOINTS" default:"^/$,^/browse/?$,^/talk/?$,^/p2p/?$,^/chat/?$,^/text2image/?$,^/tts/?$,^/static/.*$,^/swagger.*$" help:"If LOCALAI_DISABLE_API_KEY_REQUIREMENT_FOR_HTTP_GET is overriden to true, this is the list of endpoints to exempt. Only adjust this in case of a security incident or as a result of a personal security posture review" group:"hardening"`
 	Peer2Peer                          bool     `env:"LOCALAI_P2P,P2P" name:"p2p" default:"false" help:"Enable P2P mode" group:"p2p"`
 	Peer2PeerDHTInterval               int      `env:"LOCALAI_P2P_DHT_INTERVAL,P2P_DHT_INTERVAL" default:"360" name:"p2p-dht-interval" help:"Interval for DHT refresh (used during token generation)" group:"p2p"`
 	Peer2PeerOTPInterval               int      `env:"LOCALAI_P2P_OTP_INTERVAL,P2P_OTP_INTERVAL" default:"9000" name:"p2p-otp-interval" help:"Interval for OTP refresh (used during token generation)" group:"p2p"`
@@ -83,7 +83,6 @@ type RunCMD struct {
 	EnableTracing                      bool     `env:"LOCALAI_ENABLE_TRACING,ENABLE_TRACING" help:"Enable API tracing" group:"api"`
 	TracingMaxItems                    int      `env:"LOCALAI_TRACING_MAX_ITEMS" default:"1024" help:"Maximum number of traces to keep" group:"api"`
 	AgentJobRetentionDays              int      `env:"LOCALAI_AGENT_JOB_RETENTION_DAYS,AGENT_JOB_RETENTION_DAYS" default:"30" help:"Number of days to keep agent job history (default: 30)" group:"api"`
-	OpenResponsesStoreTTL               string   `env:"LOCALAI_OPEN_RESPONSES_STORE_TTL,OPEN_RESPONSES_STORE_TTL" default:"0" help:"TTL for Open Responses store (e.g., 1h, 30m, 0 = no expiration)" group:"api"`

 	Version bool
 }
@@ -250,15 +249,6 @@ func (r *RunCMD) Run(ctx *cliContext.Context) error {
 		opts = append(opts, config.WithLRUEvictionRetryInterval(dur))
 	}

-	// Handle Open Responses store TTL
-	if r.OpenResponsesStoreTTL != "" && r.OpenResponsesStoreTTL != "0" {
-		dur, err := time.ParseDuration(r.OpenResponsesStoreTTL)
-		if err != nil {
-			return fmt.Errorf("invalid Open Responses store TTL: %w", err)
-		}
-		opts = append(opts, config.WithOpenResponsesStoreTTL(dur))
-	}
-
 	// split ":" to get backend name and the uri
 	for _, v := range r.ExternalGRPCBackends {
 		backend := v[:strings.IndexByte(v, ':')]
--- a/core/config/application_config.go
+++ b/core/config/application_config.go
@@ -86,8 +86,6 @@ type ApplicationConfig struct {

 	AgentJobRetentionDays int // Default: 30 days

-	OpenResponsesStoreTTL time.Duration // TTL for Open Responses store (0 = no expiration)
-
 	PathWithoutAuth []string
 }

@@ -469,12 +467,6 @@ func WithAgentJobRetentionDays(days int) AppOption {
 	}
 }

-func WithOpenResponsesStoreTTL(ttl time.Duration) AppOption {
-	return func(o *ApplicationConfig) {
-		o.OpenResponsesStoreTTL = ttl
-	}
-}
-
 func WithEnforcedPredownloadScans(enforced bool) AppOption {
 	return func(o *ApplicationConfig) {
 		o.EnforcePredownloadScans = enforced
@@ -602,12 +594,6 @@ func (o *ApplicationConfig) ToRuntimeSettings() RuntimeSettings {
 	} else {
 		lruEvictionRetryInterval = "1s" // default
 	}
-	var openResponsesStoreTTL string
-	if o.OpenResponsesStoreTTL > 0 {
-		openResponsesStoreTTL = o.OpenResponsesStoreTTL.String()
-	} else {
-		openResponsesStoreTTL = "0" // default: no expiration
-	}

 	return RuntimeSettings{
 		WatchdogEnabled:          &watchdogEnabled,
@@ -642,7 +628,6 @@ func (o *ApplicationConfig) ToRuntimeSettings() RuntimeSettings {
 		AutoloadBackendGalleries: &autoloadBackendGalleries,
 		ApiKeys:                  &apiKeys,
 		AgentJobRetentionDays:    &agentJobRetentionDays,
-		OpenResponsesStoreTTL:     &openResponsesStoreTTL,
 	}
 }

@@ -784,14 +769,6 @@ func (o *ApplicationConfig) ApplyRuntimeSettings(settings *RuntimeSettings) (req
 	if settings.AgentJobRetentionDays != nil {
 		o.AgentJobRetentionDays = *settings.AgentJobRetentionDays
 	}
-	if settings.OpenResponsesStoreTTL != nil {
-		if *settings.OpenResponsesStoreTTL == "0" || *settings.OpenResponsesStoreTTL == "" {
-			o.OpenResponsesStoreTTL = 0 // No expiration
-		} else if dur, err := time.ParseDuration(*settings.OpenResponsesStoreTTL); err == nil {
-			o.OpenResponsesStoreTTL = dur
-		}
-		// This setting doesn't require restart, can be updated dynamically
-	}
 	// Note: ApiKeys requires special handling (merging with startup keys) - handled in caller

 	return requireRestart
--- a/core/config/gguf.go
+++ b/core/config/gguf.go
@@ -1,16 +1,10 @@
 package config

 import (
-	"context"
-
-	"github.com/mudler/LocalAI/pkg/grpc"
-	pb "github.com/mudler/LocalAI/pkg/grpc/proto"
-	"github.com/mudler/LocalAI/pkg/reasoning"
 	"github.com/mudler/LocalAI/pkg/xsysinfo"
 	"github.com/mudler/xlog"

 	gguf "github.com/gpustack/gguf-parser-go"
-	"github.com/gpustack/gguf-parser-go/util/ptr"
 )

 const (
@@ -68,25 +62,16 @@ func guessGGUFFromFile(cfg *ModelConfig, f *gguf.GGUFFile, defaultCtx int) {
 		cfg.NGPULayers = &defaultHigh
 	}

-	xlog.Debug("[gguf] guessDefaultsFromFile: NGPULayers set", "NGPULayers", cfg.NGPULayers, "modelName", f.Metadata().Name)
-
-	// identify from well known templates first, otherwise use the raw jinja template
-	chatTemplate, found := f.Header.MetadataKV.Get("tokenizer.chat_template")
-	if found {
-		// fill jinja template
-		cfg.modelTemplate = chatTemplate.ValueString()
-	}
-
-	// Thinking support detection is done after model load via DetectThinkingSupportFromBackend
+	xlog.Debug("guessDefaultsFromFile: NGPULayers set", "NGPULayers", cfg.NGPULayers)

 	// template estimations
 	if cfg.HasTemplate() {
 		// nothing to guess here
-		xlog.Debug("[gguf] guessDefaultsFromFile: template already set", "name", cfg.Name, "modelName", f.Metadata().Name)
+		xlog.Debug("guessDefaultsFromFile: template already set", "name", cfg.Name)
 		return
 	}

-	xlog.Debug("[gguf] Model file loaded", "file", cfg.ModelFileName(), "eosTokenID", f.Tokenizer().EOSTokenID, "bosTokenID", f.Tokenizer().BOSTokenID, "modelName", f.Metadata().Name, "architecture", f.Architecture().Architecture)
+	xlog.Debug("Model file loaded", "file", cfg.ModelFileName(), "eosTokenID", f.Tokenizer().EOSTokenID, "bosTokenID", f.Tokenizer().BOSTokenID, "modelName", f.Metadata().Name, "architecture", f.Architecture().Architecture)

 	// guess the name
 	if cfg.Name == "" {
@@ -98,49 +83,4 @@ func guessGGUFFromFile(cfg *ModelConfig, f *gguf.GGUFFile, defaultCtx int) {
 	cfg.FunctionsConfig.GrammarConfig.NoGrammar = true
 	cfg.Options = append(cfg.Options, "use_jinja:true")
 	cfg.KnownUsecaseStrings = append(cfg.KnownUsecaseStrings, "FLAG_CHAT")
-
-}
-
-// DetectThinkingSupportFromBackend calls the ModelMetadata gRPC method to detect
-// if the model supports thinking mode and if the template ends with a thinking start token.
-// This should be called after the model is loaded.
-// The results are stored in cfg.SupportsThinking and cfg.ThinkingForcedOpen.
-func DetectThinkingSupportFromBackend(ctx context.Context, cfg *ModelConfig, backendClient grpc.Backend, modelOptions *pb.ModelOptions) {
-	if backendClient == nil {
-		xlog.Debug("[gguf] DetectThinkingSupportFromBackend: backend client is nil, skipping detection")
-		return
-	}
-
-	if modelOptions == nil {
-		xlog.Debug("[gguf] DetectThinkingSupportFromBackend: model options is nil, skipping detection")
-		return
-	}
-
-	// Only detect for llama-cpp backend when using tokenizer templates
-	if cfg.Backend != "llama-cpp" || !cfg.TemplateConfig.UseTokenizerTemplate {
-		xlog.Debug("[gguf] DetectThinkingSupportFromBackend: skipping detection", "backend", cfg.Backend, "useTokenizerTemplate", cfg.TemplateConfig.UseTokenizerTemplate)
-		return
-	}
-
-	metadata, err := backendClient.ModelMetadata(ctx, modelOptions)
-	if err != nil {
-		xlog.Warn("[gguf] DetectThinkingSupportFromBackend: failed to get model metadata", "error", err)
-		return
-	}
-
-	if metadata != nil {
-		cfg.ReasoningConfig.DisableReasoning = ptr.To(!metadata.SupportsThinking)
-
-		// Use the rendered template to detect if thinking token is at the end
-		// This reuses the existing DetectThinkingStartToken function
-		if metadata.RenderedTemplate != "" {
-			thinkingStartToken := reasoning.DetectThinkingStartToken(metadata.RenderedTemplate, &cfg.ReasoningConfig)
-			thinkingForcedOpen := thinkingStartToken != ""
-			cfg.ReasoningConfig.DisableReasoningTagPrefill = ptr.To(!thinkingForcedOpen)
-			xlog.Debug("[gguf] DetectThinkingSupportFromBackend: thinking support detected", "supports_thinking", metadata.SupportsThinking, "thinking_forced_open", thinkingForcedOpen, "thinking_start_token", thinkingStartToken)
-		} else {
-			cfg.ReasoningConfig.DisableReasoningTagPrefill = ptr.To(true)
-			xlog.Debug("[gguf] DetectThinkingSupportFromBackend: thinking support detected", "supports_thinking", metadata.SupportsThinking, "thinking_forced_open", false)
-		}
-	}
 }
--- a/core/config/model_config.go
+++ b/core/config/model_config.go
@@ -10,7 +10,6 @@ import (
 	"github.com/mudler/LocalAI/core/schema"
 	"github.com/mudler/LocalAI/pkg/downloader"
 	"github.com/mudler/LocalAI/pkg/functions"
-	"github.com/mudler/LocalAI/pkg/reasoning"
 	"github.com/mudler/cogito"
 	"gopkg.in/yaml.v3"
 )
@@ -31,7 +30,6 @@ type TTSConfig struct {
 // @Description ModelConfig represents a model configuration
 type ModelConfig struct {
 	modelConfigFile          string `yaml:"-" json:"-"`
-	modelTemplate            string `yaml:"-" json:"-"`
 	schema.PredictionOptions `yaml:"parameters,omitempty" json:"parameters,omitempty"`
 	Name                     string `yaml:"name,omitempty" json:"name,omitempty"`

@@ -53,7 +51,6 @@ type ModelConfig struct {
 	ResponseFormatMap                          map[string]interface{} `yaml:"-" json:"-"`

 	FunctionsConfig functions.FunctionsConfig `yaml:"function,omitempty" json:"function,omitempty"`
-	ReasoningConfig reasoning.Config          `yaml:"reasoning,omitempty" json:"reasoning,omitempty"`

 	FeatureFlag FeatureFlag `yaml:"feature_flags,omitempty" json:"feature_flags,omitempty"` // Feature Flag registry. We move fast, and features may break on a per model/backend basis. Registry for (usually temporary) flags that indicate aborting something early.
 	// LLM configs (GPT4ALL, Llama.cpp, ...)
@@ -524,11 +521,6 @@ func (c *ModelConfig) GetModelConfigFile() string {
 	return c.modelConfigFile
 }

-// GetModelTemplate returns the model's chat template if available
-func (c *ModelConfig) GetModelTemplate() string {
-	return c.modelTemplate
-}
-
 type ModelConfigUsecase int

 const (
--- a/core/config/model_config_loader.go
+++ b/core/config/model_config_loader.go
@@ -246,17 +246,6 @@ func (bcl *ModelConfigLoader) RemoveModelConfig(m string) {
 	delete(bcl.configs, m)
 }

-// UpdateModelConfig updates an existing model config in the loader.
-// This is useful for updating runtime-detected properties like thinking support.
-func (bcl *ModelConfigLoader) UpdateModelConfig(m string, updater func(*ModelConfig)) {
-	bcl.Lock()
-	defer bcl.Unlock()
-	if cfg, exists := bcl.configs[m]; exists {
-		updater(&cfg)
-		bcl.configs[m] = cfg
-	}
-}
-
 // Preload prepare models if they are not local but url or huggingface repositories
 func (bcl *ModelConfigLoader) Preload(modelPath string) error {
 	bcl.Lock()
--- a/core/config/runtime_settings.go
+++ b/core/config/runtime_settings.go
@@ -60,7 +60,4 @@ type RuntimeSettings struct {

 	// Agent settings
 	AgentJobRetentionDays *int `json:"agent_job_retention_days,omitempty"`
-
-	// Open Responses settings
-	OpenResponsesStoreTTL *string `json:"open_responses_store_ttl,omitempty"` // TTL for stored responses (e.g., "1h", "30m", "0" = no expiration)
 }
--- a/core/http/app.go
+++ b/core/http/app.go
@@ -108,15 +108,7 @@ func API(application *application.Application) (*echo.Echo, error) {
 			req := c.Request()
 			res := c.Response()
 			err := next(c)
-
-			// Fix for #7989: Reduce log verbosity of Web UI polling
-			// If the path is /api/operations and the request was successful (200),
-			// we log it at DEBUG level (hidden by default) instead of INFO.
-			if req.URL.Path == "/api/operations" && res.Status == 200 {
-				xlog.Debug("HTTP request", "method", req.Method, "path", req.URL.Path, "status", res.Status)
-			} else {
-				xlog.Info("HTTP request", "method", req.Method, "path", req.URL.Path, "status", res.Status)
-			}
+			xlog.Info("HTTP request", "method", req.Method, "path", req.URL.Path, "status", res.Status)
 			return err
 		}
 	})
@@ -193,8 +185,6 @@ func API(application *application.Application) (*echo.Echo, error) {
 			corsConfig.AllowOrigins = strings.Split(application.ApplicationConfig().CORSAllowOrigins, ",")
 		}
 		e.Use(middleware.CORSWithConfig(corsConfig))
-	} else {
-		e.Use(middleware.CORS())
 	}

 	// CSRF middleware
@@ -216,7 +206,6 @@ func API(application *application.Application) (*echo.Echo, error) {
 	routes.RegisterLocalAIRoutes(e, requestExtractor, application.ModelConfigLoader(), application.ModelLoader(), application.ApplicationConfig(), application.GalleryService(), opcache, application.TemplatesEvaluator(), application)
 	routes.RegisterOpenAIRoutes(e, requestExtractor, application)
 	routes.RegisterAnthropicRoutes(e, requestExtractor, application)
-	routes.RegisterOpenResponsesRoutes(e, requestExtractor, application)
 	if !application.ApplicationConfig().DisableWebUI {
 		routes.RegisterUIAPIRoutes(e, application.ModelConfigLoader(), application.ModelLoader(), application.ApplicationConfig(), application.GalleryService(), opcache, application)
 		routes.RegisterUIRoutes(e, application.ModelConfigLoader(), application.ModelLoader(), application.ApplicationConfig(), application.GalleryService())
--- a/core/http/endpoints/localai/settings.go
+++ b/core/http/endpoints/localai/settings.go
@@ -11,7 +11,6 @@ import (
 	"github.com/labstack/echo/v4"
 	"github.com/mudler/LocalAI/core/application"
 	"github.com/mudler/LocalAI/core/config"
-	"github.com/mudler/LocalAI/core/http/endpoints/openresponses"
 	"github.com/mudler/LocalAI/core/p2p"
 	"github.com/mudler/LocalAI/core/schema"
 	"github.com/mudler/xlog"
@@ -85,16 +84,6 @@ func UpdateSettingsEndpoint(app *application.Application) echo.HandlerFunc {
 				})
 			}
 		}
-		if settings.OpenResponsesStoreTTL != nil {
-			if *settings.OpenResponsesStoreTTL != "0" && *settings.OpenResponsesStoreTTL != "" {
-				if _, err := time.ParseDuration(*settings.OpenResponsesStoreTTL); err != nil {
-					return c.JSON(http.StatusBadRequest, schema.SettingsResponse{
-						Success: false,
-						Error:   "Invalid open_responses_store_ttl format: " + err.Error(),
-					})
-				}
-			}
-		}

 		// Save to file
 		if appConfig.DynamicConfigsDir == "" {
@@ -155,22 +144,6 @@ func UpdateSettingsEndpoint(app *application.Application) echo.HandlerFunc {
 			xlog.Info("Updated LRU eviction retry settings", "maxRetries", maxRetries, "retryInterval", retryInterval)
 		}

-		// Update Open Responses store TTL dynamically
-		if settings.OpenResponsesStoreTTL != nil {
-			ttl := time.Duration(0)
-			if *settings.OpenResponsesStoreTTL != "0" && *settings.OpenResponsesStoreTTL != "" {
-				if dur, err := time.ParseDuration(*settings.OpenResponsesStoreTTL); err == nil {
-					ttl = dur
-				} else {
-					xlog.Warn("Invalid Open Responses store TTL format", "ttl", *settings.OpenResponsesStoreTTL, "error", err)
-				}
-			}
-			// Import the store package
-			store := openresponses.GetGlobalStore()
-			store.SetTTL(ttl)
-			xlog.Info("Updated Open Responses store TTL", "ttl", ttl)
-		}
-
 		// Check if agent job retention changed
 		agentJobChanged := settings.AgentJobRetentionDays != nil

--- a/core/http/endpoints/localai/video.go
+++ b/core/http/endpoints/localai/video.go
@@ -167,16 +167,6 @@ func VideoEndpoint(cl *config.ModelConfigLoader, ml *model.ModelLoader, appConfi

 		baseURL := middleware.BaseURL(c)

-		xlog.Debug("VideoEndpoint: Calling VideoGeneration",
-			"num_frames", input.NumFrames,
-			"fps", input.FPS,
-			"cfg_scale", input.CFGScale,
-			"step", input.Step,
-			"seed", input.Seed,
-			"width", width,
-			"height", height,
-			"negative_prompt", input.NegativePrompt)
-
 		fn, err := backend.VideoGeneration(
 			height,
 			width,
--- a/core/http/endpoints/localai/welcome.go
+++ b/core/http/endpoints/localai/welcome.go
@@ -65,13 +65,9 @@ func WelcomeEndpoint(appConfig *config.ApplicationConfig,
 			// The client expects a JSON response
 			return c.JSON(200, summary)
 		} else {
-			// Check if this is the manage route
-			templateName := "views/index"
-			if strings.HasSuffix(c.Request().URL.Path, "/manage") || c.Request().URL.Path == "/manage" {
-				templateName = "views/manage"
-			}
-			// Render appropriate template
-			return c.Render(200, templateName, summary)
+			// Serve the SPA for both index and manage routes
+			// The SPA handles routing client-side via Alpine.js
+			return c.Render(200, "views/spa", summary)
 		}
 	}
 }
--- a/core/http/endpoints/openai/chat.go
+++ b/core/http/endpoints/openai/chat.go
@@ -13,7 +13,6 @@ import (
 	"github.com/mudler/LocalAI/core/http/middleware"
 	"github.com/mudler/LocalAI/core/schema"
 	"github.com/mudler/LocalAI/pkg/functions"
-	reason "github.com/mudler/LocalAI/pkg/reasoning"

 	"github.com/mudler/LocalAI/core/templates"
 	"github.com/mudler/LocalAI/pkg/model"
@@ -39,16 +38,6 @@ func ChatEndpoint(cl *config.ModelConfigLoader, ml *model.ModelLoader, evaluator
 		}
 		responses <- initialMessage

-		// Detect if thinking token is already in prompt or template
-		// When UseTokenizerTemplate is enabled, predInput is empty, so we check the template
-		var template string
-		if config.TemplateConfig.UseTokenizerTemplate {
-			template = config.GetModelTemplate()
-		} else {
-			template = s
-		}
-		thinkingStartToken := reason.DetectThinkingStartToken(template, &config.ReasoningConfig)
-
 		// Track accumulated content for reasoning extraction
 		accumulatedContent := ""
 		lastEmittedReasoning := ""
@@ -56,8 +45,8 @@ func ChatEndpoint(cl *config.ModelConfigLoader, ml *model.ModelLoader, evaluator

 		_, _, err := ComputeChoices(req, s, config, cl, startupOptions, loader, func(s string, c *[]schema.Choice) {}, func(s string, tokenUsage backend.TokenUsage) bool {
 			accumulatedContent += s
-
-			currentReasoning, cleanedContent := reason.ExtractReasoningWithConfig(accumulatedContent, thinkingStartToken, config.ReasoningConfig)
+			// Extract reasoning from accumulated content
+			currentReasoning, cleanedContent := functions.ExtractReasoning(accumulatedContent)

 			// Calculate new reasoning delta (what we haven't emitted yet)
 			var reasoningDelta *string
@@ -129,15 +118,6 @@ func ChatEndpoint(cl *config.ModelConfigLoader, ml *model.ModelLoader, evaluator
 		return err
 	}
 	processTools := func(noAction string, prompt string, req *schema.OpenAIRequest, config *config.ModelConfig, loader *model.ModelLoader, responses chan schema.OpenAIResponse, extraUsage bool) error {
-		// Detect if thinking token is already in prompt or template
-		var template string
-		if config.TemplateConfig.UseTokenizerTemplate {
-			template = config.GetModelTemplate()
-		} else {
-			template = prompt
-		}
-		thinkingStartToken := reason.DetectThinkingStartToken(template, &config.ReasoningConfig)
-
 		result := ""
 		lastEmittedCount := 0
 		_, tokenUsage, err := ComputeChoices(req, prompt, config, cl, startupOptions, loader, func(s string, c *[]schema.Choice) {}, func(s string, usage backend.TokenUsage) bool {
@@ -249,8 +229,9 @@ func ChatEndpoint(cl *config.ModelConfigLoader, ml *model.ModelLoader, evaluator
 		if err != nil {
 			return err
 		}
-		// Prepend thinking token if needed, then extract reasoning before processing tool calls
-		reasoning, result := reason.ExtractReasoningWithConfig(result, thinkingStartToken, config.ReasoningConfig)
+		// Extract reasoning before processing tool calls
+		reasoning, cleanedResult := functions.ExtractReasoning(result)
+		result = cleanedResult

 		textContentToReturn = functions.ParseTextContent(result, config.FunctionsConfig)
 		result = functions.CleanupLLMResult(result, config.FunctionsConfig)
@@ -636,20 +617,11 @@ func ChatEndpoint(cl *config.ModelConfigLoader, ml *model.ModelLoader, evaluator

 		// no streaming mode
 		default:
-			// Detect if thinking token is already in prompt or template
-			var template string
-			if config.TemplateConfig.UseTokenizerTemplate {
-				template = config.GetModelTemplate() // TODO: this should be the parsed jinja template. But for now this is the best we can do.
-			} else {
-				template = predInput
-			}
-			thinkingStartToken := reason.DetectThinkingStartToken(template, &config.ReasoningConfig)
-
-			xlog.Debug("Thinking start token", "thinkingStartToken", thinkingStartToken, "template", template)

 			tokenCallback := func(s string, c *[]schema.Choice) {
-				// Prepend thinking token if needed, then extract reasoning from the response
-				reasoning, s := reason.ExtractReasoningWithConfig(s, thinkingStartToken, config.ReasoningConfig)
+				// Extract reasoning from the response
+				reasoning, cleanedS := functions.ExtractReasoning(s)
+				s = cleanedS

 				if !shouldUseFn {
 					// no function is called, just reply and use stop as finish reason
--- a/core/http/endpoints/openai/video.go
+++ b/core/http/endpoints/openai/video.go
@@ -0,0 +1,140 @@
+package openai
+
+import (
+	"encoding/json"
+	"fmt"
+	"strconv"
+	"strings"
+
+	"github.com/labstack/echo/v4"
+	"github.com/mudler/LocalAI/core/config"
+	"github.com/mudler/LocalAI/core/http/endpoints/localai"
+	"github.com/mudler/LocalAI/core/http/middleware"
+	"github.com/mudler/LocalAI/core/schema"
+	model "github.com/mudler/LocalAI/pkg/model"
+)
+
+func VideoEndpoint(cl *config.ModelConfigLoader, ml *model.ModelLoader, appConfig *config.ApplicationConfig) echo.HandlerFunc {
+	return func(c echo.Context) error {
+		input, ok := c.Get(middleware.CONTEXT_LOCALS_KEY_LOCALAI_REQUEST).(*schema.OpenAIRequest)
+		if !ok || input == nil {
+			return echo.ErrBadRequest
+		}
+		var raw map[string]interface{}
+		body := make([]byte, 0)
+		if c.Request().Body != nil {
+			c.Request().Body.Read(body)
+		}
+		if len(body) > 0 {
+			_ = json.Unmarshal(body, &raw)
+		}
+		// Build VideoRequest using shared mapper
+		vr := MapOpenAIToVideo(input, raw)
+		// Place VideoRequest into context so localai.VideoEndpoint can consume it
+		c.Set(middleware.CONTEXT_LOCALS_KEY_LOCALAI_REQUEST, vr)
+		// Delegate to existing localai handler
+		return localai.VideoEndpoint(cl, ml, appConfig)(c)
+	}
+}
+
+// VideoEndpoint godoc
+// @Summary Generate a video from an OpenAI-compatible request
+// @Description Accepts an OpenAI-style request and delegates to the LocalAI video generator
+// @Tags openai
+// @Accept json
+// @Produce json
+// @Param request body schema.OpenAIRequest true "OpenAI-style request"
+// @Success 200 {object} map[string]interface{}
+// @Failure 400 {object} map[string]interface{}
+// @Router /v1/videos [post]
+
+func MapOpenAIToVideo(input *schema.OpenAIRequest, raw map[string]interface{}) *schema.VideoRequest {
+	vr := &schema.VideoRequest{}
+	if input == nil {
+		return vr
+	}
+
+	if input.Model != "" {
+		vr.Model = input.Model
+	}
+
+	// Prompt mapping
+	switch p := input.Prompt.(type) {
+	case string:
+		vr.Prompt = p
+	case []interface{}:
+		if len(p) > 0 {
+			if s, ok := p[0].(string); ok {
+				vr.Prompt = s
+			}
+		}
+	}
+
+	// Size
+	size := input.Size
+	if size == "" && raw != nil {
+		if v, ok := raw["size"].(string); ok {
+			size = v
+		}
+	}
+	if size != "" {
+		parts := strings.SplitN(size, "x", 2)
+		if len(parts) == 2 {
+			if wi, err := strconv.Atoi(parts[0]); err == nil {
+				vr.Width = int32(wi)
+			}
+			if hi, err := strconv.Atoi(parts[1]); err == nil {
+				vr.Height = int32(hi)
+			}
+		}
+	}
+
+	// seconds -> num frames
+	secondsStr := ""
+	if raw != nil {
+		if v, ok := raw["seconds"].(string); ok {
+			secondsStr = v
+		} else if v, ok := raw["seconds"].(float64); ok {
+			secondsStr = fmt.Sprintf("%v", int(v))
+		}
+	}
+	fps := int32(30)
+	if raw != nil {
+		if rawFPS, ok := raw["fps"]; ok {
+			switch rf := rawFPS.(type) {
+			case float64:
+				fps = int32(rf)
+			case string:
+				if fi, err := strconv.Atoi(rf); err == nil {
+					fps = int32(fi)
+				}
+			}
+		}
+	}
+	if secondsStr != "" {
+		if secF, err := strconv.Atoi(secondsStr); err == nil {
+			vr.FPS = fps
+			vr.NumFrames = int32(secF) * fps
+		}
+	}
+
+	// input_reference
+	if raw != nil {
+		if v, ok := raw["input_reference"].(string); ok {
+			vr.StartImage = v
+		}
+	}
+
+	// response format
+	if input.ResponseFormat != nil {
+		if rf, ok := input.ResponseFormat.(string); ok {
+			vr.ResponseFormat = rf
+		}
+	}
+
+	if input.Step != 0 {
+		vr.Step = int32(input.Step)
+	}
+
+	return vr
+}
--- a/core/http/endpoints/openresponses/responses.go
+++ b/core/http/endpoints/openresponses/responses.go
--- a/core/http/endpoints/openresponses/store.go
+++ b/core/http/endpoints/openresponses/store.go
@@ -1,453 +0,0 @@
-package openresponses
-
-import (
-	"context"
-	"encoding/json"
-	"fmt"
-	"sync"
-	"time"
-
-	"github.com/mudler/LocalAI/core/schema"
-	"github.com/mudler/xlog"
-)
-
-// ResponseStore provides thread-safe storage for Open Responses API responses
-type ResponseStore struct {
-	mu            sync.RWMutex
-	responses     map[string]*StoredResponse
-	ttl           time.Duration // Time-to-live for stored responses (0 = no expiration)
-	cleanupCtx    context.Context
-	cleanupCancel context.CancelFunc
-}
-
-// StreamedEvent represents a buffered SSE event for streaming resume
-type StreamedEvent struct {
-	SequenceNumber int    `json:"sequence_number"`
-	EventType      string `json:"event_type"`
-	Data           []byte `json:"data"` // JSON-serialized event
-}
-
-// StoredResponse contains a complete response with its input request and output items
-type StoredResponse struct {
-	Request   *schema.OpenResponsesRequest
-	Response  *schema.ORResponseResource
-	Items     map[string]*schema.ORItemField // item_id -> item mapping for quick lookup
-	StoredAt  time.Time
-	ExpiresAt *time.Time // nil if no expiration
-
-	// Background execution support
-	CancelFunc    context.CancelFunc // For cancellation of background tasks
-	StreamEvents  []StreamedEvent    // Buffered events for streaming resume
-	StreamEnabled bool               // Was created with stream=true
-	IsBackground  bool               // Was created with background=true
-	EventsChan    chan struct{}      // Signals new events for live subscribers
-	mu            sync.RWMutex       // Protect concurrent access to this response
-}
-
-var (
-	globalStore *ResponseStore
-	storeOnce   sync.Once
-)
-
-// GetGlobalStore returns the singleton response store instance
-func GetGlobalStore() *ResponseStore {
-	storeOnce.Do(func() {
-		globalStore = NewResponseStore(0) // Default: no TTL, will be updated from appConfig
-	})
-	return globalStore
-}
-
-// SetTTL updates the TTL for the store
-// This will affect all new responses stored after this call
-func (s *ResponseStore) SetTTL(ttl time.Duration) {
-	s.mu.Lock()
-	defer s.mu.Unlock()
-
-	// Stop existing cleanup loop if running
-	if s.cleanupCancel != nil {
-		s.cleanupCancel()
-		s.cleanupCancel = nil
-		s.cleanupCtx = nil
-	}
-
-	s.ttl = ttl
-
-	// If TTL > 0, start cleanup loop
-	if ttl > 0 {
-		s.cleanupCtx, s.cleanupCancel = context.WithCancel(context.Background())
-		go s.cleanupLoop(s.cleanupCtx)
-	}
-
-	xlog.Debug("Updated Open Responses store TTL", "ttl", ttl, "cleanup_running", ttl > 0)
-}
-
-// NewResponseStore creates a new response store with optional TTL
-// If ttl is 0, responses are stored indefinitely
-func NewResponseStore(ttl time.Duration) *ResponseStore {
-	store := &ResponseStore{
-		responses: make(map[string]*StoredResponse),
-		ttl:       ttl,
-	}
-
-	// Start cleanup goroutine if TTL is set
-	if ttl > 0 {
-		store.cleanupCtx, store.cleanupCancel = context.WithCancel(context.Background())
-		go store.cleanupLoop(store.cleanupCtx)
-	}
-
-	return store
-}
-
-// Store stores a response with its request and items
-func (s *ResponseStore) Store(responseID string, request *schema.OpenResponsesRequest, response *schema.ORResponseResource) {
-	s.mu.Lock()
-	defer s.mu.Unlock()
-
-	// Build item index for quick lookup
-	items := make(map[string]*schema.ORItemField)
-	for i := range response.Output {
-		item := &response.Output[i]
-		if item.ID != "" {
-			items[item.ID] = item
-		}
-	}
-
-	stored := &StoredResponse{
-		Request:   request,
-		Response:  response,
-		Items:     items,
-		StoredAt:  time.Now(),
-		ExpiresAt: nil,
-	}
-
-	// Set expiration if TTL is configured
-	if s.ttl > 0 {
-		expiresAt := time.Now().Add(s.ttl)
-		stored.ExpiresAt = &expiresAt
-	}
-
-	s.responses[responseID] = stored
-	xlog.Debug("Stored Open Responses response", "response_id", responseID, "items_count", len(items))
-}
-
-// Get retrieves a stored response by ID
-func (s *ResponseStore) Get(responseID string) (*StoredResponse, error) {
-	s.mu.RLock()
-	defer s.mu.RUnlock()
-
-	stored, exists := s.responses[responseID]
-	if !exists {
-		return nil, fmt.Errorf("response not found: %s", responseID)
-	}
-
-	// Check expiration
-	if stored.ExpiresAt != nil && time.Now().After(*stored.ExpiresAt) {
-		// Expired, but we'll return it anyway and let caller handle cleanup
-		return nil, fmt.Errorf("response expired: %s", responseID)
-	}
-
-	return stored, nil
-}
-
-// GetItem retrieves a specific item from a stored response
-func (s *ResponseStore) GetItem(responseID, itemID string) (*schema.ORItemField, error) {
-	stored, err := s.Get(responseID)
-	if err != nil {
-		return nil, err
-	}
-
-	item, exists := stored.Items[itemID]
-	if !exists {
-		return nil, fmt.Errorf("item not found: %s in response %s", itemID, responseID)
-	}
-
-	return item, nil
-}
-
-// FindItem searches for an item across all stored responses
-// Returns the item and the response ID it was found in
-func (s *ResponseStore) FindItem(itemID string) (*schema.ORItemField, string, error) {
-	s.mu.RLock()
-	defer s.mu.RUnlock()
-
-	now := time.Now()
-	for responseID, stored := range s.responses {
-		// Skip expired responses
-		if stored.ExpiresAt != nil && now.After(*stored.ExpiresAt) {
-			continue
-		}
-
-		if item, exists := stored.Items[itemID]; exists {
-			return item, responseID, nil
-		}
-	}
-
-	return nil, "", fmt.Errorf("item not found in any stored response: %s", itemID)
-}
-
-// Delete removes a response from storage
-func (s *ResponseStore) Delete(responseID string) {
-	s.mu.Lock()
-	defer s.mu.Unlock()
-	delete(s.responses, responseID)
-	xlog.Debug("Deleted Open Responses response", "response_id", responseID)
-}
-
-// Cleanup removes expired responses
-func (s *ResponseStore) Cleanup() int {
-	if s.ttl == 0 {
-		return 0
-	}
-
-	s.mu.Lock()
-	defer s.mu.Unlock()
-
-	now := time.Now()
-	count := 0
-	for id, stored := range s.responses {
-		if stored.ExpiresAt != nil && now.After(*stored.ExpiresAt) {
-			delete(s.responses, id)
-			count++
-		}
-	}
-
-	if count > 0 {
-		xlog.Debug("Cleaned up expired Open Responses", "count", count)
-	}
-
-	return count
-}
-
-// cleanupLoop runs periodic cleanup of expired responses
-func (s *ResponseStore) cleanupLoop(ctx context.Context) {
-	if s.ttl == 0 {
-		return
-	}
-
-	ticker := time.NewTicker(s.ttl / 2) // Cleanup at half TTL interval
-	defer ticker.Stop()
-
-	for {
-		select {
-		case <-ctx.Done():
-			xlog.Debug("Stopped Open Responses store cleanup loop")
-			return
-		case <-ticker.C:
-			s.Cleanup()
-		}
-	}
-}
-
-// Count returns the number of stored responses
-func (s *ResponseStore) Count() int {
-	s.mu.RLock()
-	defer s.mu.RUnlock()
-	return len(s.responses)
-}
-
-// StoreBackground stores a background response with cancel function and optional streaming support
-func (s *ResponseStore) StoreBackground(responseID string, request *schema.OpenResponsesRequest, response *schema.ORResponseResource, cancelFunc context.CancelFunc, streamEnabled bool) {
-	s.mu.Lock()
-	defer s.mu.Unlock()
-
-	// Build item index for quick lookup
-	items := make(map[string]*schema.ORItemField)
-	for i := range response.Output {
-		item := &response.Output[i]
-		if item.ID != "" {
-			items[item.ID] = item
-		}
-	}
-
-	stored := &StoredResponse{
-		Request:       request,
-		Response:      response,
-		Items:         items,
-		StoredAt:      time.Now(),
-		ExpiresAt:     nil,
-		CancelFunc:    cancelFunc,
-		StreamEvents:  []StreamedEvent{},
-		StreamEnabled: streamEnabled,
-		IsBackground:  true,
-		EventsChan:    make(chan struct{}, 100), // Buffered channel for event notifications
-	}
-
-	// Set expiration if TTL is configured
-	if s.ttl > 0 {
-		expiresAt := time.Now().Add(s.ttl)
-		stored.ExpiresAt = &expiresAt
-	}
-
-	s.responses[responseID] = stored
-	xlog.Debug("Stored background Open Responses response", "response_id", responseID, "stream_enabled", streamEnabled)
-}
-
-// UpdateStatus updates the status of a stored response
-func (s *ResponseStore) UpdateStatus(responseID string, status string, completedAt *int64) error {
-	s.mu.RLock()
-	stored, exists := s.responses[responseID]
-	s.mu.RUnlock()
-
-	if !exists {
-		return fmt.Errorf("response not found: %s", responseID)
-	}
-
-	stored.mu.Lock()
-	defer stored.mu.Unlock()
-
-	stored.Response.Status = status
-	stored.Response.CompletedAt = completedAt
-
-	xlog.Debug("Updated response status", "response_id", responseID, "status", status)
-	return nil
-}
-
-// UpdateResponse updates the entire response object for a stored response
-func (s *ResponseStore) UpdateResponse(responseID string, response *schema.ORResponseResource) error {
-	s.mu.RLock()
-	stored, exists := s.responses[responseID]
-	s.mu.RUnlock()
-
-	if !exists {
-		return fmt.Errorf("response not found: %s", responseID)
-	}
-
-	stored.mu.Lock()
-	defer stored.mu.Unlock()
-
-	// Rebuild item index
-	items := make(map[string]*schema.ORItemField)
-	for i := range response.Output {
-		item := &response.Output[i]
-		if item.ID != "" {
-			items[item.ID] = item
-		}
-	}
-
-	stored.Response = response
-	stored.Items = items
-
-	xlog.Debug("Updated response", "response_id", responseID, "status", response.Status, "items_count", len(items))
-	return nil
-}
-
-// AppendEvent appends a streaming event to the buffer for resume support
-func (s *ResponseStore) AppendEvent(responseID string, event *schema.ORStreamEvent) error {
-	s.mu.RLock()
-	stored, exists := s.responses[responseID]
-	s.mu.RUnlock()
-
-	if !exists {
-		return fmt.Errorf("response not found: %s", responseID)
-	}
-
-	// Serialize the event
-	data, err := json.Marshal(event)
-	if err != nil {
-		return fmt.Errorf("failed to marshal event: %w", err)
-	}
-
-	stored.mu.Lock()
-	stored.StreamEvents = append(stored.StreamEvents, StreamedEvent{
-		SequenceNumber: event.SequenceNumber,
-		EventType:      event.Type,
-		Data:           data,
-	})
-	stored.mu.Unlock()
-
-	// Notify any subscribers of new event
-	select {
-	case stored.EventsChan <- struct{}{}:
-	default:
-		// Channel full, subscribers will catch up
-	}
-
-	return nil
-}
-
-// GetEventsAfter returns all events with sequence number greater than startingAfter
-func (s *ResponseStore) GetEventsAfter(responseID string, startingAfter int) ([]StreamedEvent, error) {
-	s.mu.RLock()
-	stored, exists := s.responses[responseID]
-	s.mu.RUnlock()
-
-	if !exists {
-		return nil, fmt.Errorf("response not found: %s", responseID)
-	}
-
-	stored.mu.RLock()
-	defer stored.mu.RUnlock()
-
-	var result []StreamedEvent
-	for _, event := range stored.StreamEvents {
-		if event.SequenceNumber > startingAfter {
-			result = append(result, event)
-		}
-	}
-
-	return result, nil
-}
-
-// Cancel cancels a background response if it's still in progress
-func (s *ResponseStore) Cancel(responseID string) (*schema.ORResponseResource, error) {
-	s.mu.RLock()
-	stored, exists := s.responses[responseID]
-	s.mu.RUnlock()
-
-	if !exists {
-		return nil, fmt.Errorf("response not found: %s", responseID)
-	}
-
-	stored.mu.Lock()
-	defer stored.mu.Unlock()
-
-	// If already in a terminal state, just return the response (idempotent)
-	status := stored.Response.Status
-	if status == schema.ORStatusCompleted || status == schema.ORStatusFailed ||
-		status == schema.ORStatusIncomplete || status == schema.ORStatusCancelled {
-		xlog.Debug("Response already in terminal state", "response_id", responseID, "status", status)
-		return stored.Response, nil
-	}
-
-	// Cancel the context if available
-	if stored.CancelFunc != nil {
-		stored.CancelFunc()
-		xlog.Debug("Cancelled background response", "response_id", responseID)
-	}
-
-	// Update status to cancelled
-	now := time.Now().Unix()
-	stored.Response.Status = schema.ORStatusCancelled
-	stored.Response.CompletedAt = &now
-
-	return stored.Response, nil
-}
-
-// GetEventsChan returns the events notification channel for a response
-func (s *ResponseStore) GetEventsChan(responseID string) (chan struct{}, error) {
-	s.mu.RLock()
-	stored, exists := s.responses[responseID]
-	s.mu.RUnlock()
-
-	if !exists {
-		return nil, fmt.Errorf("response not found: %s", responseID)
-	}
-
-	return stored.EventsChan, nil
-}
-
-// IsStreamEnabled checks if a response was created with streaming enabled
-func (s *ResponseStore) IsStreamEnabled(responseID string) (bool, error) {
-	s.mu.RLock()
-	stored, exists := s.responses[responseID]
-	s.mu.RUnlock()
-
-	if !exists {
-		return false, fmt.Errorf("response not found: %s", responseID)
-	}
-
-	stored.mu.RLock()
-	defer stored.mu.RUnlock()
-
-	return stored.StreamEnabled, nil
-}
--- a/core/http/endpoints/openresponses/store_suite_test.go
+++ b/core/http/endpoints/openresponses/store_suite_test.go
@@ -1,13 +0,0 @@
-package openresponses
-
-import (
-	"testing"
-
-	. "github.com/onsi/ginkgo/v2"
-	. "github.com/onsi/gomega"
-)
-
-func TestStore(t *testing.T) {
-	RegisterFailHandler(Fail)
-	RunSpecs(t, "ResponseStore Suite")
-}
--- a/core/http/endpoints/openresponses/store_test.go
+++ b/core/http/endpoints/openresponses/store_test.go
@@ -1,626 +0,0 @@
-package openresponses
-
-import (
-	"context"
-	"fmt"
-	"time"
-
-	"github.com/mudler/LocalAI/core/schema"
-	. "github.com/onsi/ginkgo/v2"
-	. "github.com/onsi/gomega"
-)
-
-var _ = Describe("ResponseStore", func() {
-	var store *ResponseStore
-
-	BeforeEach(func() {
-		store = NewResponseStore(0) // No TTL for most tests
-	})
-
-	AfterEach(func() {
-		// Clean up
-	})
-
-	Describe("Store and Get", func() {
-		It("should store and retrieve a response", func() {
-			responseID := "resp_test123"
-			request := &schema.OpenResponsesRequest{
-				Model: "test-model",
-				Input: "Hello",
-			}
-			response := &schema.ORResponseResource{
-				ID:        responseID,
-				Object:    "response",
-				CreatedAt: time.Now().Unix(),
-				Status:    "completed",
-				Model:     "test-model",
-				Output: []schema.ORItemField{
-					{
-						Type:   "message",
-						ID:     "msg_123",
-						Status: "completed",
-						Role:   "assistant",
-						Content: []schema.ORContentPart{{
-							Type:        "output_text",
-							Text:        "Hello, world!",
-							Annotations: []schema.ORAnnotation{},
-							Logprobs:    []schema.ORLogProb{},
-						}},
-					},
-				},
-			}
-
-			store.Store(responseID, request, response)
-
-			stored, err := store.Get(responseID)
-			Expect(err).ToNot(HaveOccurred())
-			Expect(stored).ToNot(BeNil())
-			Expect(stored.Response.ID).To(Equal(responseID))
-			Expect(stored.Request.Model).To(Equal("test-model"))
-			Expect(len(stored.Items)).To(Equal(1))
-			Expect(stored.Items["msg_123"]).ToNot(BeNil())
-			Expect(stored.Items["msg_123"].ID).To(Equal("msg_123"))
-		})
-
-		It("should return error for non-existent response", func() {
-			_, err := store.Get("nonexistent")
-			Expect(err).To(HaveOccurred())
-			Expect(err.Error()).To(ContainSubstring("not found"))
-		})
-
-		It("should index all items by ID", func() {
-			responseID := "resp_test456"
-			request := &schema.OpenResponsesRequest{
-				Model: "test-model",
-				Input: "Test",
-			}
-			response := &schema.ORResponseResource{
-				ID:     responseID,
-				Object: "response",
-				Output: []schema.ORItemField{
-					{
-						Type:   "message",
-						ID:     "msg_1",
-						Status: "completed",
-						Role:   "assistant",
-					},
-					{
-						Type:      "function_call",
-						ID:        "fc_1",
-						Status:    "completed",
-						CallID:    "fc_1",
-						Name:      "test_function",
-						Arguments: `{"arg": "value"}`,
-					},
-					{
-						Type:   "message",
-						ID:     "msg_2",
-						Status: "completed",
-						Role:   "assistant",
-					},
-				},
-			}
-
-			store.Store(responseID, request, response)
-
-			stored, err := store.Get(responseID)
-			Expect(err).ToNot(HaveOccurred())
-			Expect(len(stored.Items)).To(Equal(3))
-			Expect(stored.Items["msg_1"]).ToNot(BeNil())
-			Expect(stored.Items["fc_1"]).ToNot(BeNil())
-			Expect(stored.Items["msg_2"]).ToNot(BeNil())
-		})
-
-		It("should handle items without IDs", func() {
-			responseID := "resp_test789"
-			request := &schema.OpenResponsesRequest{
-				Model: "test-model",
-				Input: "Test",
-			}
-			response := &schema.ORResponseResource{
-				ID:     responseID,
-				Object: "response",
-				Output: []schema.ORItemField{
-					{
-						Type:   "message",
-						ID:     "", // No ID
-						Status: "completed",
-						Role:   "assistant",
-					},
-					{
-						Type:   "message",
-						ID:     "msg_with_id",
-						Status: "completed",
-						Role:   "assistant",
-					},
-				},
-			}
-
-			store.Store(responseID, request, response)
-
-			stored, err := store.Get(responseID)
-			Expect(err).ToNot(HaveOccurred())
-			// Only items with IDs are indexed
-			Expect(len(stored.Items)).To(Equal(1))
-			Expect(stored.Items["msg_with_id"]).ToNot(BeNil())
-		})
-	})
-
-	Describe("GetItem", func() {
-		It("should retrieve a specific item by ID", func() {
-			responseID := "resp_item_test"
-			itemID := "msg_specific"
-			request := &schema.OpenResponsesRequest{
-				Model: "test-model",
-				Input: "Test",
-			}
-			response := &schema.ORResponseResource{
-				ID:     responseID,
-				Object: "response",
-				Output: []schema.ORItemField{
-					{
-						Type:   "message",
-						ID:     itemID,
-						Status: "completed",
-						Role:   "assistant",
-						Content: []schema.ORContentPart{{
-							Type:        "output_text",
-							Text:        "Specific message",
-							Annotations: []schema.ORAnnotation{},
-							Logprobs:    []schema.ORLogProb{},
-						}},
-					},
-				},
-			}
-
-			store.Store(responseID, request, response)
-
-			item, err := store.GetItem(responseID, itemID)
-			Expect(err).ToNot(HaveOccurred())
-			Expect(item).ToNot(BeNil())
-			Expect(item.ID).To(Equal(itemID))
-			Expect(item.Type).To(Equal("message"))
-		})
-
-		It("should return error for non-existent item", func() {
-			responseID := "resp_item_test2"
-			request := &schema.OpenResponsesRequest{
-				Model: "test-model",
-				Input: "Test",
-			}
-			response := &schema.ORResponseResource{
-				ID:     responseID,
-				Object: "response",
-				Output: []schema.ORItemField{
-					{
-						Type:   "message",
-						ID:     "msg_existing",
-						Status: "completed",
-					},
-				},
-			}
-
-			store.Store(responseID, request, response)
-
-			_, err := store.GetItem(responseID, "nonexistent_item")
-			Expect(err).To(HaveOccurred())
-			Expect(err.Error()).To(ContainSubstring("item not found"))
-		})
-
-		It("should return error for non-existent response when getting item", func() {
-			_, err := store.GetItem("nonexistent_response", "any_item")
-			Expect(err).To(HaveOccurred())
-			Expect(err.Error()).To(ContainSubstring("response not found"))
-		})
-	})
-
-	Describe("FindItem", func() {
-		It("should find an item across all stored responses", func() {
-			// Store first response
-			responseID1 := "resp_find_1"
-			itemID1 := "msg_find_1"
-			store.Store(responseID1, &schema.OpenResponsesRequest{Model: "test"}, &schema.ORResponseResource{
-				ID:     responseID1,
-				Object: "response",
-				Output: []schema.ORItemField{
-					{Type: "message", ID: itemID1, Status: "completed"},
-				},
-			})
-
-			// Store second response
-			responseID2 := "resp_find_2"
-			itemID2 := "msg_find_2"
-			store.Store(responseID2, &schema.OpenResponsesRequest{Model: "test"}, &schema.ORResponseResource{
-				ID:     responseID2,
-				Object: "response",
-				Output: []schema.ORItemField{
-					{Type: "message", ID: itemID2, Status: "completed"},
-				},
-			})
-
-			// Find item from first response
-			item, foundResponseID, err := store.FindItem(itemID1)
-			Expect(err).ToNot(HaveOccurred())
-			Expect(item).ToNot(BeNil())
-			Expect(item.ID).To(Equal(itemID1))
-			Expect(foundResponseID).To(Equal(responseID1))
-
-			// Find item from second response
-			item, foundResponseID, err = store.FindItem(itemID2)
-			Expect(err).ToNot(HaveOccurred())
-			Expect(item).ToNot(BeNil())
-			Expect(item.ID).To(Equal(itemID2))
-			Expect(foundResponseID).To(Equal(responseID2))
-		})
-
-		It("should return error when item not found in any response", func() {
-			_, _, err := store.FindItem("nonexistent_item")
-			Expect(err).To(HaveOccurred())
-			Expect(err.Error()).To(ContainSubstring("item not found in any stored response"))
-		})
-	})
-
-	Describe("Delete", func() {
-		It("should delete a stored response", func() {
-			responseID := "resp_delete_test"
-			request := &schema.OpenResponsesRequest{Model: "test"}
-			response := &schema.ORResponseResource{
-				ID:     responseID,
-				Object: "response",
-			}
-
-			store.Store(responseID, request, response)
-			Expect(store.Count()).To(Equal(1))
-
-			store.Delete(responseID)
-			Expect(store.Count()).To(Equal(0))
-
-			_, err := store.Get(responseID)
-			Expect(err).To(HaveOccurred())
-		})
-
-		It("should handle deleting non-existent response gracefully", func() {
-			// Should not panic
-			store.Delete("nonexistent")
-			Expect(store.Count()).To(Equal(0))
-		})
-	})
-
-	Describe("Count", func() {
-		It("should return correct count of stored responses", func() {
-			Expect(store.Count()).To(Equal(0))
-
-			store.Store("resp_1", &schema.OpenResponsesRequest{Model: "test"}, &schema.ORResponseResource{ID: "resp_1", Object: "response"})
-			Expect(store.Count()).To(Equal(1))
-
-			store.Store("resp_2", &schema.OpenResponsesRequest{Model: "test"}, &schema.ORResponseResource{ID: "resp_2", Object: "response"})
-			Expect(store.Count()).To(Equal(2))
-
-			store.Delete("resp_1")
-			Expect(store.Count()).To(Equal(1))
-		})
-	})
-
-	Describe("TTL and Expiration", func() {
-		It("should set expiration when TTL is configured", func() {
-			ttlStore := NewResponseStore(100 * time.Millisecond)
-			responseID := "resp_ttl_test"
-			request := &schema.OpenResponsesRequest{Model: "test"}
-			response := &schema.ORResponseResource{ID: responseID, Object: "response"}
-
-			ttlStore.Store(responseID, request, response)
-
-			stored, err := ttlStore.Get(responseID)
-			Expect(err).ToNot(HaveOccurred())
-			Expect(stored.ExpiresAt).ToNot(BeNil())
-			Expect(stored.ExpiresAt.After(time.Now())).To(BeTrue())
-		})
-
-		It("should not set expiration when TTL is 0", func() {
-			responseID := "resp_no_ttl"
-			request := &schema.OpenResponsesRequest{Model: "test"}
-			response := &schema.ORResponseResource{ID: responseID, Object: "response"}
-
-			store.Store(responseID, request, response)
-
-			stored, err := store.Get(responseID)
-			Expect(err).ToNot(HaveOccurred())
-			Expect(stored.ExpiresAt).To(BeNil())
-		})
-
-		It("should clean up expired responses", func() {
-			ttlStore := NewResponseStore(50 * time.Millisecond)
-			responseID := "resp_expire_test"
-			request := &schema.OpenResponsesRequest{Model: "test"}
-			response := &schema.ORResponseResource{ID: responseID, Object: "response"}
-
-			ttlStore.Store(responseID, request, response)
-			Expect(ttlStore.Count()).To(Equal(1))
-
-			// Wait for expiration (longer than TTL and cleanup interval)
-			time.Sleep(150 * time.Millisecond)
-
-			// Cleanup should remove expired response (may have already been cleaned by goroutine)
-			count := ttlStore.Cleanup()
-			// Count might be 0 if cleanup goroutine already ran, or 1 if we're first
-			Expect(count).To(BeNumerically(">=", 0))
-			Expect(ttlStore.Count()).To(Equal(0))
-
-			_, err := ttlStore.Get(responseID)
-			Expect(err).To(HaveOccurred())
-		})
-
-		It("should return error for expired response", func() {
-			ttlStore := NewResponseStore(50 * time.Millisecond)
-			responseID := "resp_expire_error"
-			request := &schema.OpenResponsesRequest{Model: "test"}
-			response := &schema.ORResponseResource{ID: responseID, Object: "response"}
-
-			ttlStore.Store(responseID, request, response)
-
-			// Wait for expiration (but not long enough for cleanup goroutine to remove it)
-			time.Sleep(75 * time.Millisecond)
-
-			// Try to get before cleanup goroutine removes it
-			_, err := ttlStore.Get(responseID)
-			// Error could be "expired" or "not found" (if cleanup already ran)
-			Expect(err).To(HaveOccurred())
-			// Either error message is acceptable
-			errMsg := err.Error()
-			Expect(errMsg).To(Or(ContainSubstring("expired"), ContainSubstring("not found")))
-		})
-	})
-
-	Describe("Thread Safety", func() {
-		It("should handle concurrent stores and gets", func() {
-			// This is a basic concurrency test
-			done := make(chan bool, 10)
-			for i := 0; i < 10; i++ {
-				go func(id int) {
-					responseID := fmt.Sprintf("resp_concurrent_%d", id)
-					request := &schema.OpenResponsesRequest{Model: "test"}
-					response := &schema.ORResponseResource{
-						ID:     responseID,
-						Object: "response",
-						Output: []schema.ORItemField{
-							{Type: "message", ID: fmt.Sprintf("msg_%d", id), Status: "completed"},
-						},
-					}
-					store.Store(responseID, request, response)
-
-					// Retrieve immediately
-					stored, err := store.Get(responseID)
-					Expect(err).ToNot(HaveOccurred())
-					Expect(stored).ToNot(BeNil())
-					done <- true
-				}(i)
-			}
-
-			// Wait for all goroutines
-			for i := 0; i < 10; i++ {
-				<-done
-			}
-
-			Expect(store.Count()).To(Equal(10))
-		})
-	})
-
-	Describe("GetGlobalStore", func() {
-		It("should return singleton instance", func() {
-			store1 := GetGlobalStore()
-			store2 := GetGlobalStore()
-			Expect(store1).To(Equal(store2))
-		})
-
-		It("should persist data across GetGlobalStore calls", func() {
-			globalStore := GetGlobalStore()
-			responseID := "resp_global_test"
-			request := &schema.OpenResponsesRequest{Model: "test"}
-			response := &schema.ORResponseResource{ID: responseID, Object: "response"}
-
-			globalStore.Store(responseID, request, response)
-
-			// Get store again
-			globalStore2 := GetGlobalStore()
-			stored, err := globalStore2.Get(responseID)
-			Expect(err).ToNot(HaveOccurred())
-			Expect(stored).ToNot(BeNil())
-		})
-	})
-
-	Describe("Background Mode Support", func() {
-		It("should store background response with cancel function", func() {
-			responseID := "resp_bg_test"
-			request := &schema.OpenResponsesRequest{Model: "test"}
-			response := &schema.ORResponseResource{
-				ID:     responseID,
-				Object: "response",
-				Status: schema.ORStatusQueued,
-			}
-
-			_, cancel := context.WithCancel(context.Background())
-			defer cancel()
-
-			store.StoreBackground(responseID, request, response, cancel, true)
-
-			stored, err := store.Get(responseID)
-			Expect(err).ToNot(HaveOccurred())
-			Expect(stored).ToNot(BeNil())
-			Expect(stored.IsBackground).To(BeTrue())
-			Expect(stored.StreamEnabled).To(BeTrue())
-			Expect(stored.CancelFunc).ToNot(BeNil())
-		})
-
-		It("should update status of stored response", func() {
-			responseID := "resp_status_test"
-			request := &schema.OpenResponsesRequest{Model: "test"}
-			response := &schema.ORResponseResource{
-				ID:     responseID,
-				Object: "response",
-				Status: schema.ORStatusQueued,
-			}
-
-			store.Store(responseID, request, response)
-
-			err := store.UpdateStatus(responseID, schema.ORStatusInProgress, nil)
-			Expect(err).ToNot(HaveOccurred())
-
-			stored, err := store.Get(responseID)
-			Expect(err).ToNot(HaveOccurred())
-			Expect(stored.Response.Status).To(Equal(schema.ORStatusInProgress))
-		})
-
-		It("should append and retrieve streaming events", func() {
-			responseID := "resp_events_test"
-			request := &schema.OpenResponsesRequest{Model: "test"}
-			response := &schema.ORResponseResource{
-				ID:     responseID,
-				Object: "response",
-				Status: schema.ORStatusInProgress,
-			}
-
-			_, cancel := context.WithCancel(context.Background())
-			defer cancel()
-
-			store.StoreBackground(responseID, request, response, cancel, true)
-
-			// Append events
-			event1 := &schema.ORStreamEvent{
-				Type:           "response.created",
-				SequenceNumber: 0,
-			}
-			event2 := &schema.ORStreamEvent{
-				Type:           "response.in_progress",
-				SequenceNumber: 1,
-			}
-			event3 := &schema.ORStreamEvent{
-				Type:           "response.output_text.delta",
-				SequenceNumber: 2,
-			}
-
-			err := store.AppendEvent(responseID, event1)
-			Expect(err).ToNot(HaveOccurred())
-			err = store.AppendEvent(responseID, event2)
-			Expect(err).ToNot(HaveOccurred())
-			err = store.AppendEvent(responseID, event3)
-			Expect(err).ToNot(HaveOccurred())
-
-			// Get all events after -1 (all events)
-			events, err := store.GetEventsAfter(responseID, -1)
-			Expect(err).ToNot(HaveOccurred())
-			Expect(events).To(HaveLen(3))
-
-			// Get events after sequence 1
-			events, err = store.GetEventsAfter(responseID, 1)
-			Expect(err).ToNot(HaveOccurred())
-			Expect(events).To(HaveLen(1))
-			Expect(events[0].SequenceNumber).To(Equal(2))
-		})
-
-		It("should cancel an in-progress response", func() {
-			responseID := "resp_cancel_test"
-			request := &schema.OpenResponsesRequest{Model: "test"}
-			response := &schema.ORResponseResource{
-				ID:     responseID,
-				Object: "response",
-				Status: schema.ORStatusInProgress,
-			}
-
-			_, cancel := context.WithCancel(context.Background())
-			defer cancel()
-
-			store.StoreBackground(responseID, request, response, cancel, false)
-
-			// Cancel the response
-			cancelledResponse, err := store.Cancel(responseID)
-			Expect(err).ToNot(HaveOccurred())
-			Expect(cancelledResponse.Status).To(Equal(schema.ORStatusCancelled))
-			Expect(cancelledResponse.CompletedAt).ToNot(BeNil())
-		})
-
-		It("should be idempotent when cancelling already completed response", func() {
-			responseID := "resp_idempotent_cancel"
-			request := &schema.OpenResponsesRequest{Model: "test"}
-			completedAt := time.Now().Unix()
-			response := &schema.ORResponseResource{
-				ID:          responseID,
-				Object:      "response",
-				Status:      schema.ORStatusCompleted,
-				CompletedAt: &completedAt,
-			}
-
-			store.Store(responseID, request, response)
-
-			// Try to cancel a completed response
-			cancelledResponse, err := store.Cancel(responseID)
-			Expect(err).ToNot(HaveOccurred())
-			// Status should remain completed (not changed to cancelled)
-			Expect(cancelledResponse.Status).To(Equal(schema.ORStatusCompleted))
-		})
-
-		It("should check if streaming is enabled", func() {
-			responseID := "resp_stream_check"
-			request := &schema.OpenResponsesRequest{Model: "test"}
-			response := &schema.ORResponseResource{
-				ID:     responseID,
-				Object: "response",
-				Status: schema.ORStatusQueued,
-			}
-
-			_, cancel := context.WithCancel(context.Background())
-			defer cancel()
-
-			store.StoreBackground(responseID, request, response, cancel, true)
-
-			enabled, err := store.IsStreamEnabled(responseID)
-			Expect(err).ToNot(HaveOccurred())
-			Expect(enabled).To(BeTrue())
-
-			// Store another without streaming
-			responseID2 := "resp_no_stream"
-			store.StoreBackground(responseID2, request, response, cancel, false)
-
-			enabled2, err := store.IsStreamEnabled(responseID2)
-			Expect(err).ToNot(HaveOccurred())
-			Expect(enabled2).To(BeFalse())
-		})
-
-		It("should notify subscribers of new events", func() {
-			responseID := "resp_events_chan"
-			request := &schema.OpenResponsesRequest{Model: "test"}
-			response := &schema.ORResponseResource{
-				ID:     responseID,
-				Object: "response",
-				Status: schema.ORStatusInProgress,
-			}
-
-			_, cancel := context.WithCancel(context.Background())
-			defer cancel()
-
-			store.StoreBackground(responseID, request, response, cancel, true)
-
-			eventsChan, err := store.GetEventsChan(responseID)
-			Expect(err).ToNot(HaveOccurred())
-			Expect(eventsChan).ToNot(BeNil())
-
-			// Append an event
-			event := &schema.ORStreamEvent{
-				Type:           "response.output_text.delta",
-				SequenceNumber: 0,
-			}
-
-			go func() {
-				time.Sleep(10 * time.Millisecond)
-				store.AppendEvent(responseID, event)
-			}()
-
-			// Wait for notification
-			select {
-			case <-eventsChan:
-				// Event received
-			case <-time.After(1 * time.Second):
-				Fail("Timeout waiting for event notification")
-			}
-		})
-	})
-})
--- a/core/http/http_suite_test.go
+++ b/core/http/http_suite_test.go
@@ -1,33 +1,13 @@
 package http_test

 import (
-	"os"
-	"path/filepath"
 	"testing"

 	. "github.com/onsi/ginkgo/v2"
 	. "github.com/onsi/gomega"
 )

-var (
-	tmpdir   string
-	modelDir string
-)
-
 func TestLocalAI(t *testing.T) {
 	RegisterFailHandler(Fail)
-
-	var err error
-	tmpdir, err = os.MkdirTemp("", "")
-	Expect(err).ToNot(HaveOccurred())
-	modelDir = filepath.Join(tmpdir, "models")
-	err = os.Mkdir(modelDir, 0750)
-	Expect(err).ToNot(HaveOccurred())
-
-	AfterSuite(func() {
-		err := os.RemoveAll(tmpdir)
-		Expect(err).ToNot(HaveOccurred())
-	})
-
 	RunSpecs(t, "LocalAI HTTP test suite")
 }
--- a/core/http/middleware/request.go
+++ b/core/http/middleware/request.go
@@ -484,103 +484,3 @@ func mergeOpenAIRequestAndModelConfig(config *config.ModelConfig, input *schema.
 	}
 	return fmt.Errorf("unable to validate configuration after merging")
 }
-
-func (re *RequestExtractor) SetOpenResponsesRequest(c echo.Context) error {
-	input, ok := c.Get(CONTEXT_LOCALS_KEY_LOCALAI_REQUEST).(*schema.OpenResponsesRequest)
-	if !ok || input.Model == "" {
-		return echo.ErrBadRequest
-	}
-
-	cfg, ok := c.Get(CONTEXT_LOCALS_KEY_MODEL_CONFIG).(*config.ModelConfig)
-	if !ok || cfg == nil {
-		return echo.ErrBadRequest
-	}
-
-	// Extract or generate the correlation ID (Open Responses uses x-request-id)
-	correlationID := c.Request().Header.Get("x-request-id")
-	if correlationID == "" {
-		correlationID = uuid.New().String()
-	}
-	c.Response().Header().Set("x-request-id", correlationID)
-
-	// Use the request context directly - Echo properly supports context cancellation!
-	reqCtx := c.Request().Context()
-	c1, cancel := context.WithCancel(re.applicationConfig.Context)
-
-	// Cancel when request context is cancelled (client disconnects)
-	go func() {
-		select {
-		case <-reqCtx.Done():
-			cancel()
-		case <-c1.Done():
-			// Already cancelled
-		}
-	}()
-
-	// Add the correlation ID to the new context
-	ctxWithCorrelationID := context.WithValue(c1, CorrelationIDKey, correlationID)
-
-	input.Context = ctxWithCorrelationID
-	input.Cancel = cancel
-
-	err := mergeOpenResponsesRequestAndModelConfig(cfg, input)
-	if err != nil {
-		return err
-	}
-
-	if cfg.Model == "" {
-		xlog.Debug("replacing empty cfg.Model with input value", "input.Model", input.Model)
-		cfg.Model = input.Model
-	}
-
-	c.Set(CONTEXT_LOCALS_KEY_LOCALAI_REQUEST, input)
-	c.Set(CONTEXT_LOCALS_KEY_MODEL_CONFIG, cfg)
-
-	return nil
-}
-
-func mergeOpenResponsesRequestAndModelConfig(config *config.ModelConfig, input *schema.OpenResponsesRequest) error {
-	// Temperature
-	if input.Temperature != nil {
-		config.Temperature = input.Temperature
-	}
-
-	// TopP
-	if input.TopP != nil {
-		config.TopP = input.TopP
-	}
-
-	// MaxOutputTokens -> Maxtokens
-	if input.MaxOutputTokens != nil {
-		config.Maxtokens = input.MaxOutputTokens
-	}
-
-	// Convert tools to functions - this will be handled in the endpoint handler
-	// We just validate that tools are present if needed
-
-	// Handle tool_choice
-	if input.ToolChoice != nil {
-		switch tc := input.ToolChoice.(type) {
-		case string:
-			// "auto", "required", or "none"
-			if tc == "required" {
-				config.SetFunctionCallString("required")
-			} else if tc == "none" {
-				// Don't use tools - handled in endpoint
-			}
-			// "auto" is default - let model decide
-		case map[string]interface{}:
-			// Specific tool: {type:"function", name:"..."}
-			if tcType, ok := tc["type"].(string); ok && tcType == "function" {
-				if name, ok := tc["name"].(string); ok {
-					config.SetFunctionCallString(name)
-				}
-			}
-		}
-	}
-
-	if valid, _ := config.Validate(); valid {
-		return nil
-	}
-	return fmt.Errorf("unable to validate configuration after merging")
-}
--- a/core/http/middleware/trace.go
+++ b/core/http/middleware/trace.go
@@ -36,7 +36,6 @@ type APIExchange struct {
 var traceBuffer *circularbuffer.Queue[APIExchange]
 var mu sync.Mutex
 var logChan = make(chan APIExchange, 100)
-var initOnce sync.Once

 type bodyWriter struct {
 	http.ResponseWriter
@@ -54,37 +53,26 @@ func (w *bodyWriter) Flush() {
 	}
 }

-func initializeTracing(maxItems int) {
-	initOnce.Do(func() {
-		if maxItems <= 0 {
-			maxItems = 100
-		}
-		mu.Lock()
-		traceBuffer = circularbuffer.New[APIExchange](maxItems)
-		mu.Unlock()
+// TraceMiddleware intercepts and logs JSON API requests and responses
+func TraceMiddleware(app *application.Application) echo.MiddlewareFunc {
+	if app.ApplicationConfig().EnableTracing && traceBuffer == nil {
+		traceBuffer = circularbuffer.New[APIExchange](app.ApplicationConfig().TracingMaxItems)

 		go func() {
 			for exchange := range logChan {
 				mu.Lock()
-				if traceBuffer != nil {
-					traceBuffer.Enqueue(exchange)
-				}
+				traceBuffer.Enqueue(exchange)
 				mu.Unlock()
 			}
 		}()
-	})
-}
+	}

-// TraceMiddleware intercepts and logs JSON API requests and responses
-func TraceMiddleware(app *application.Application) echo.MiddlewareFunc {
 	return func(next echo.HandlerFunc) echo.HandlerFunc {
 		return func(c echo.Context) error {
 			if !app.ApplicationConfig().EnableTracing {
 				return next(c)
 			}

-			initializeTracing(app.ApplicationConfig().TracingMaxItems)
-
 			if c.Request().Header.Get("Content-Type") != "application/json" {
 				return next(c)
 			}
@@ -150,10 +138,6 @@ func TraceMiddleware(app *application.Application) echo.MiddlewareFunc {
 // GetTraces returns a copy of the logged API exchanges for display
 func GetTraces() []APIExchange {
 	mu.Lock()
-	if traceBuffer == nil {
-		mu.Unlock()
-		return []APIExchange{}
-	}
 	traces := traceBuffer.Values()
 	mu.Unlock()

@@ -167,8 +151,6 @@ func GetTraces() []APIExchange {
 // ClearTraces clears the in-memory logs
 func ClearTraces() {
 	mu.Lock()
-	if traceBuffer != nil {
-		traceBuffer.Clear()
-	}
+	traceBuffer.Clear()
 	mu.Unlock()
 }
--- a/core/http/openai_mapping_test.go
+++ b/core/http/openai_mapping_test.go
@@ -0,0 +1,75 @@
+package http_test
+
+import (
+	"encoding/json"
+
+	openai "github.com/mudler/LocalAI/core/http/endpoints/openai"
+	"github.com/mudler/LocalAI/core/schema"
+
+	. "github.com/onsi/ginkgo/v2"
+	. "github.com/onsi/gomega"
+)
+
+var _ = Describe("MapOpenAIToVideo", func() {
+	It("maps size and seconds correctly", func() {
+		cases := []struct {
+			name     string
+			input    *schema.OpenAIRequest
+			raw      map[string]interface{}
+			expectsW int32
+			expectsH int32
+			expectsF int32
+			expectsN int32
+		}{
+			{
+				name: "size in input",
+				input: &schema.OpenAIRequest{
+					PredictionOptions: schema.PredictionOptions{
+						BasicModelRequest: schema.BasicModelRequest{Model: "m"},
+					},
+					Size: "256x128",
+				},
+				expectsW: 256,
+				expectsH: 128,
+			},
+			{
+				name:  "size in raw and seconds as string",
+				input: &schema.OpenAIRequest{PredictionOptions: schema.PredictionOptions{BasicModelRequest: schema.BasicModelRequest{Model: "m"}}},
+				raw:   map[string]interface{}{"size": "720x480", "seconds": "2"},
+				expectsW: 720,
+				expectsH: 480,
+				expectsF: 30,
+				expectsN: 60,
+			},
+			{
+				name:  "seconds as number and fps override",
+				input: &schema.OpenAIRequest{PredictionOptions: schema.PredictionOptions{BasicModelRequest: schema.BasicModelRequest{Model: "m"}}},
+				raw:   map[string]interface{}{"seconds": 3.0, "fps": 24.0},
+				expectsF: 24,
+				expectsN: 72,
+			},
+		}
+
+		for _, c := range cases {
+			By(c.name)
+			vr := openai.MapOpenAIToVideo(c.input, c.raw)
+			if c.expectsW != 0 {
+				Expect(vr.Width).To(Equal(c.expectsW))
+			}
+			if c.expectsH != 0 {
+				Expect(vr.Height).To(Equal(c.expectsH))
+			}
+			if c.expectsF != 0 {
+				Expect(vr.FPS).To(Equal(c.expectsF))
+			}
+			if c.expectsN != 0 {
+				Expect(vr.NumFrames).To(Equal(c.expectsN))
+			}
+
+			b, err := json.Marshal(vr)
+			Expect(err).ToNot(HaveOccurred())
+			_ = b
+		}
+	})
+})
+
--- a/core/http/openai_videos_test.go
+++ b/core/http/openai_videos_test.go
@@ -0,0 +1,168 @@
+package http_test
+
+import (
+    "bytes"
+    "context"
+    "encoding/json"
+    "io"
+    "net/http"
+    "os"
+    "path/filepath"
+    "time"
+
+    "github.com/mudler/LocalAI/core/application"
+    "github.com/mudler/LocalAI/core/config"
+    "github.com/mudler/LocalAI/pkg/system"
+    "github.com/mudler/LocalAI/pkg/grpc"
+    pb "github.com/mudler/LocalAI/pkg/grpc/proto"
+    "fmt"
+    . "github.com/mudler/LocalAI/core/http"
+    "github.com/labstack/echo/v4"
+
+    . "github.com/onsi/ginkgo/v2"
+    . "github.com/onsi/gomega"
+)
+
+const testAPIKey = "joshua"
+
+type fakeAI struct{}
+
+func (f *fakeAI) Busy() bool                                { return false }
+func (f *fakeAI) Lock()                                     {}
+func (f *fakeAI) Unlock()                                   {}
+func (f *fakeAI) Locking() bool                             { return false }
+func (f *fakeAI) Predict(*pb.PredictOptions) (string, error) { return "", nil }
+func (f *fakeAI) PredictStream(*pb.PredictOptions, chan string) error {
+    return nil
+}
+func (f *fakeAI) Load(*pb.ModelOptions) error                        { return nil }
+func (f *fakeAI) Embeddings(*pb.PredictOptions) ([]float32, error)    { return nil, nil }
+func (f *fakeAI) GenerateImage(*pb.GenerateImageRequest) error       { return nil }
+func (f *fakeAI) GenerateVideo(*pb.GenerateVideoRequest) error       { return nil }
+func (f *fakeAI) Detect(*pb.DetectOptions) (pb.DetectResponse, error) { return pb.DetectResponse{}, nil }
+func (f *fakeAI) AudioTranscription(*pb.TranscriptRequest) (pb.TranscriptResult, error) {
+    return pb.TranscriptResult{}, nil
+}
+func (f *fakeAI) TTS(*pb.TTSRequest) error                          { return nil }
+func (f *fakeAI) SoundGeneration(*pb.SoundGenerationRequest) error  { return nil }
+func (f *fakeAI) TokenizeString(*pb.PredictOptions) (pb.TokenizationResponse, error) {
+    return pb.TokenizationResponse{}, nil
+}
+func (f *fakeAI) Status() (pb.StatusResponse, error) { return pb.StatusResponse{}, nil }
+func (f *fakeAI) StoresSet(*pb.StoresSetOptions) error        { return nil }
+func (f *fakeAI) StoresDelete(*pb.StoresDeleteOptions) error  { return nil }
+func (f *fakeAI) StoresGet(*pb.StoresGetOptions) (pb.StoresGetResult, error) {
+    return pb.StoresGetResult{}, nil
+}
+func (f *fakeAI) StoresFind(*pb.StoresFindOptions) (pb.StoresFindResult, error) {
+    return pb.StoresFindResult{}, nil
+}
+func (f *fakeAI) VAD(*pb.VADRequest) (pb.VADResponse, error) { return pb.VADResponse{}, nil }
+
+var _ = Describe("OpenAI /v1/videos (embedded backend)", func() {
+    var tmpdir string
+    var appServer *application.Application
+    var app *echo.Echo
+    var ctx context.Context
+    var cancel context.CancelFunc
+
+    BeforeEach(func() {
+        var err error
+        tmpdir, err = os.MkdirTemp("", "")
+        Expect(err).ToNot(HaveOccurred())
+
+        modelDir := filepath.Join(tmpdir, "models")
+        err = os.Mkdir(modelDir, 0750)
+        Expect(err).ToNot(HaveOccurred())
+
+        ctx, cancel = context.WithCancel(context.Background())
+
+        systemState, err := system.GetSystemState(
+            system.WithModelPath(modelDir),
+        )
+        Expect(err).ToNot(HaveOccurred())
+
+   		grpc.Provide("embedded://fake", &fakeAI{})
+
+        appServer, err = application.New(
+            config.WithContext(ctx),
+            config.WithSystemState(systemState),
+            config.WithApiKeys([]string{testAPIKey}),
+            config.WithGeneratedContentDir(tmpdir),
+            config.WithExternalBackend("fake", "embedded://fake"),
+        )
+        Expect(err).ToNot(HaveOccurred())
+    })
+
+    AfterEach(func() {
+        cancel()
+        if app != nil {
+            ctx, cancel := context.WithTimeout(context.Background(), 5*time.Second)
+            defer cancel()
+            _ = app.Shutdown(ctx)
+        }
+        _ = os.RemoveAll(tmpdir)
+    })
+
+    It("accepts OpenAI-style video create and delegates to backend", func() {
+		var err error
+		app, err = API(appServer)
+		Expect(err).ToNot(HaveOccurred())
+		go func() {
+			if err := app.Start("127.0.0.1:9091"); err != nil && err != http.ErrServerClosed {
+				// Log error if needed
+			}
+		}()
+
+        // wait for server
+        client := &http.Client{Timeout: 5 * time.Second}
+        Eventually(func() error {
+            req, _ := http.NewRequest("GET", "http://127.0.0.1:9091/v1/models", nil)
+            req.Header.Set("Authorization", "Bearer "+testAPIKey)
+            resp, err := client.Do(req)
+            if err != nil {
+                return err
+            }
+            defer resp.Body.Close()
+            if resp.StatusCode >= 400 {
+                return fmt.Errorf("bad status: %d", resp.StatusCode)
+            }
+            return nil
+        }, "30s", "500ms").Should(Succeed())
+
+        body := map[string]interface{}{
+            "model": "fake-model",
+            "backend": "fake",
+            "prompt": "a test video",
+            "size": "256x256",
+            "seconds": "1",
+        }
+        payload, err := json.Marshal(body)
+        Expect(err).ToNot(HaveOccurred())
+
+        req, err := http.NewRequest("POST", "http://127.0.0.1:9091/v1/videos", bytes.NewBuffer(payload))
+        Expect(err).ToNot(HaveOccurred())
+        req.Header.Set("Content-Type", "application/json")
+        req.Header.Set("Authorization", "Bearer "+testAPIKey)
+
+        resp, err := client.Do(req)
+        Expect(err).ToNot(HaveOccurred())
+        defer resp.Body.Close()
+        Expect(resp.StatusCode).To(Equal(200))
+
+        dat, err := io.ReadAll(resp.Body)
+        Expect(err).ToNot(HaveOccurred())
+
+        var out map[string]interface{}
+        err = json.Unmarshal(dat, &out)
+        Expect(err).ToNot(HaveOccurred())
+        data, ok := out["data"].([]interface{})
+        Expect(ok).To(BeTrue())
+        Expect(len(data)).To(BeNumerically(">", 0))
+        first := data[0].(map[string]interface{})
+        url, ok := first["url"].(string)
+        Expect(ok).To(BeTrue())
+        Expect(url).To(ContainSubstring("/generated-videos/"))
+        Expect(url).To(ContainSubstring(".mp4"))
+    })
+})
--- a/core/http/openresponses_test.go
+++ b/core/http/openresponses_test.go
--- a/core/http/routes/openai.go
+++ b/core/http/routes/openai.go
@@ -152,6 +152,27 @@ func RegisterOpenAIRoutes(app *echo.Echo,
 	app.POST("/v1/images/inpainting", inpaintingHandler, imageMiddleware...)
 	app.POST("/images/inpainting", inpaintingHandler, imageMiddleware...)

+	// videos (OpenAI-compatible endpoints mapped to LocalAI video handler)
+	videoHandler := openai.VideoEndpoint(application.ModelConfigLoader(), application.ModelLoader(), application.ApplicationConfig())
+	videoMiddleware := []echo.MiddlewareFunc{
+		traceMiddleware,
+		re.BuildFilteredFirstAvailableDefaultModel(config.BuildUsecaseFilterFn(config.FLAG_VIDEO)),
+		re.SetModelAndConfig(func() schema.LocalAIRequest { return new(schema.OpenAIRequest) }),
+		func(next echo.HandlerFunc) echo.HandlerFunc {
+			return func(c echo.Context) error {
+				if err := re.SetOpenAIRequest(c); err != nil {
+					return err
+				}
+				return next(c)
+			}
+		},
+	}
+
+	// OpenAI-style create video endpoint
+	app.POST("/v1/videos", videoHandler, videoMiddleware...)
+	app.POST("/v1/videos/generations", videoHandler, videoMiddleware...)
+	app.POST("/videos", videoHandler, videoMiddleware...)
+
 	// List models
 	app.GET("/v1/models", openai.ListModelsEndpoint(application.ModelConfigLoader(), application.ModelLoader(), application.ApplicationConfig()))
 	app.GET("/models", openai.ListModelsEndpoint(application.ModelConfigLoader(), application.ModelLoader(), application.ApplicationConfig()))
--- a/core/http/routes/openresponses.go
+++ b/core/http/routes/openresponses.go
@@ -1,58 +0,0 @@
-package routes
-
-import (
-	"github.com/labstack/echo/v4"
-	"github.com/mudler/LocalAI/core/application"
-	"github.com/mudler/LocalAI/core/config"
-	"github.com/mudler/LocalAI/core/http/endpoints/openresponses"
-	"github.com/mudler/LocalAI/core/http/middleware"
-	"github.com/mudler/LocalAI/core/schema"
-)
-
-func RegisterOpenResponsesRoutes(app *echo.Echo,
-	re *middleware.RequestExtractor,
-	application *application.Application) {
-
-	// Open Responses API endpoint
-	responsesHandler := openresponses.ResponsesEndpoint(
-		application.ModelConfigLoader(),
-		application.ModelLoader(),
-		application.TemplatesEvaluator(),
-		application.ApplicationConfig(),
-	)
-
-	responsesMiddleware := []echo.MiddlewareFunc{
-		middleware.TraceMiddleware(application),
-		re.BuildFilteredFirstAvailableDefaultModel(config.BuildUsecaseFilterFn(config.FLAG_CHAT)),
-		re.SetModelAndConfig(func() schema.LocalAIRequest { return new(schema.OpenResponsesRequest) }),
-		setOpenResponsesRequestContext(re),
-	}
-
-	// Main Open Responses endpoint
-	app.POST("/v1/responses", responsesHandler, responsesMiddleware...)
-
-	// Also support without version prefix for compatibility
-	app.POST("/responses", responsesHandler, responsesMiddleware...)
-
-	// GET /responses/:id - Retrieve a response (for polling background requests)
-	getResponseHandler := openresponses.GetResponseEndpoint()
-	app.GET("/v1/responses/:id", getResponseHandler, middleware.TraceMiddleware(application))
-	app.GET("/responses/:id", getResponseHandler, middleware.TraceMiddleware(application))
-
-	// POST /responses/:id/cancel - Cancel a background response
-	cancelResponseHandler := openresponses.CancelResponseEndpoint()
-	app.POST("/v1/responses/:id/cancel", cancelResponseHandler, middleware.TraceMiddleware(application))
-	app.POST("/responses/:id/cancel", cancelResponseHandler, middleware.TraceMiddleware(application))
-}
-
-// setOpenResponsesRequestContext sets up the context and cancel function for Open Responses requests
-func setOpenResponsesRequestContext(re *middleware.RequestExtractor) echo.MiddlewareFunc {
-	return func(next echo.HandlerFunc) echo.HandlerFunc {
-		return func(c echo.Context) error {
-			if err := re.SetOpenResponsesRequest(c); err != nil {
-				return err
-			}
-			return next(c)
-		}
-	}
-}
--- a/core/http/routes/ui.go
+++ b/core/http/routes/ui.go
@@ -3,7 +3,6 @@ package routes
 import (
 	"github.com/labstack/echo/v4"
 	"github.com/mudler/LocalAI/core/config"
-	"github.com/mudler/LocalAI/core/gallery"
 	"github.com/mudler/LocalAI/core/http/endpoints/localai"
 	"github.com/mudler/LocalAI/core/http/middleware"
 	"github.com/mudler/LocalAI/core/services"
@@ -115,258 +114,24 @@ func RegisterUIRoutes(app *echo.Echo,
 		registerBackendGalleryRoutes(app, appConfig, galleryService, processingOps)
 	}

-	app.GET("/talk", func(c echo.Context) error {
-		modelConfigs, _ := services.ListModels(cl, ml, config.NoFilterFn, services.SKIP_IF_CONFIGURED)
+	// Talk route - now served by SPA
+	app.GET("/talk", localai.WelcomeEndpoint(appConfig, cl, ml, processingOps))

-		if len(modelConfigs) == 0 {
-			// If no model is available redirect to the index which suggests how to install models
-			return c.Redirect(302, middleware.BaseURL(c))
-		}
+	// Chat routes - now served by SPA
+	app.GET("/chat", localai.WelcomeEndpoint(appConfig, cl, ml, processingOps))

-		summary := map[string]interface{}{
-			"Title":        "LocalAI - Talk",
-			"BaseURL":      middleware.BaseURL(c),
-			"ModelsConfig": modelConfigs,
-			"Model":        modelConfigs[0],
+	// Show the Chat page with specific model
+	app.GET("/chat/:model", localai.WelcomeEndpoint(appConfig, cl, ml, processingOps))

-			"Version": internal.PrintableVersion(),
-		}
+	// Text2Image routes - now served by SPA
+	app.GET("/text2image/:model", localai.WelcomeEndpoint(appConfig, cl, ml, processingOps))

-		// Render index
-		return c.Render(200, "views/talk", summary)
-	})
+	app.GET("/text2image", localai.WelcomeEndpoint(appConfig, cl, ml, processingOps))

-	app.GET("/chat", func(c echo.Context) error {
-		modelConfigs := cl.GetAllModelsConfigs()
-		modelsWithoutConfig, _ := services.ListModels(cl, ml, config.NoFilterFn, services.LOOSE_ONLY)
+	// TTS routes - now served by SPA
+	app.GET("/tts/:model", localai.WelcomeEndpoint(appConfig, cl, ml, processingOps))

-		if len(modelConfigs)+len(modelsWithoutConfig) == 0 {
-			// If no model is available redirect to the index which suggests how to install models
-			return c.Redirect(302, middleware.BaseURL(c))
-		}
-		modelThatCanBeUsed := ""
-		galleryConfigs := map[string]*gallery.ModelConfig{}
-
-		for _, m := range modelConfigs {
-			cfg, err := gallery.GetLocalModelConfiguration(ml.ModelPath, m.Name)
-			if err != nil {
-				continue
-			}
-			galleryConfigs[m.Name] = cfg
-		}
-
-		title := "LocalAI - Chat"
-		var modelContextSize *int
-
-		for _, b := range modelConfigs {
-			if b.HasUsecases(config.FLAG_CHAT) {
-				modelThatCanBeUsed = b.Name
-				title = "LocalAI - Chat with " + modelThatCanBeUsed
-				if b.LLMConfig.ContextSize != nil {
-					modelContextSize = b.LLMConfig.ContextSize
-				}
-				break
-			}
-		}
-
-		summary := map[string]interface{}{
-			"Title":               title,
-			"BaseURL":             middleware.BaseURL(c),
-			"ModelsWithoutConfig": modelsWithoutConfig,
-			"GalleryConfig":       galleryConfigs,
-			"ModelsConfig":        modelConfigs,
-			"Model":               modelThatCanBeUsed,
-			"ContextSize":         modelContextSize,
-			"Version":             internal.PrintableVersion(),
-		}
-
-		// Render index
-		return c.Render(200, "views/chat", summary)
-	})
-
-	// Show the Chat page
-	app.GET("/chat/:model", func(c echo.Context) error {
-		modelConfigs := cl.GetAllModelsConfigs()
-		modelsWithoutConfig, _ := services.ListModels(cl, ml, config.NoFilterFn, services.LOOSE_ONLY)
-
-		galleryConfigs := map[string]*gallery.ModelConfig{}
-		modelName := c.Param("model")
-		var modelContextSize *int
-
-		for _, m := range modelConfigs {
-			cfg, err := gallery.GetLocalModelConfiguration(ml.ModelPath, m.Name)
-			if err != nil {
-				continue
-			}
-			galleryConfigs[m.Name] = cfg
-			if m.Name == modelName && m.LLMConfig.ContextSize != nil {
-				modelContextSize = m.LLMConfig.ContextSize
-			}
-		}
-
-		summary := map[string]interface{}{
-			"Title":               "LocalAI - Chat with " + modelName,
-			"BaseURL":             middleware.BaseURL(c),
-			"ModelsConfig":        modelConfigs,
-			"GalleryConfig":       galleryConfigs,
-			"ModelsWithoutConfig": modelsWithoutConfig,
-			"Model":               modelName,
-			"ContextSize":         modelContextSize,
-			"Version":             internal.PrintableVersion(),
-		}
-
-		// Render index
-		return c.Render(200, "views/chat", summary)
-	})
-
-	app.GET("/image/:model", func(c echo.Context) error {
-		modelConfigs := cl.GetAllModelsConfigs()
-		modelsWithoutConfig, _ := services.ListModels(cl, ml, config.NoFilterFn, services.LOOSE_ONLY)
-
-		summary := map[string]interface{}{
-			"Title":               "LocalAI - Generate images with " + c.Param("model"),
-			"BaseURL":             middleware.BaseURL(c),
-			"ModelsConfig":        modelConfigs,
-			"ModelsWithoutConfig": modelsWithoutConfig,
-			"Model":               c.Param("model"),
-			"Version":             internal.PrintableVersion(),
-		}
-
-		// Render index
-		return c.Render(200, "views/image", summary)
-	})
-
-	app.GET("/image", func(c echo.Context) error {
-		modelConfigs := cl.GetAllModelsConfigs()
-		modelsWithoutConfig, _ := services.ListModels(cl, ml, config.NoFilterFn, services.LOOSE_ONLY)
-
-		if len(modelConfigs)+len(modelsWithoutConfig) == 0 {
-			// If no model is available redirect to the index which suggests how to install models
-			return c.Redirect(302, middleware.BaseURL(c))
-		}
-
-		modelThatCanBeUsed := ""
-		title := "LocalAI - Generate images"
-
-		for _, b := range modelConfigs {
-			if b.HasUsecases(config.FLAG_IMAGE) {
-				modelThatCanBeUsed = b.Name
-				title = "LocalAI - Generate images with " + modelThatCanBeUsed
-				break
-			}
-		}
-
-		summary := map[string]interface{}{
-			"Title":               title,
-			"BaseURL":             middleware.BaseURL(c),
-			"ModelsConfig":        modelConfigs,
-			"ModelsWithoutConfig": modelsWithoutConfig,
-			"Model":               modelThatCanBeUsed,
-			"Version":             internal.PrintableVersion(),
-		}
-
-		// Render index
-		return c.Render(200, "views/image", summary)
-	})
-
-	app.GET("/tts/:model", func(c echo.Context) error {
-		modelConfigs := cl.GetAllModelsConfigs()
-		modelsWithoutConfig, _ := services.ListModels(cl, ml, config.NoFilterFn, services.LOOSE_ONLY)
-
-		summary := map[string]interface{}{
-			"Title":               "LocalAI - Generate images with " + c.Param("model"),
-			"BaseURL":             middleware.BaseURL(c),
-			"ModelsConfig":        modelConfigs,
-			"ModelsWithoutConfig": modelsWithoutConfig,
-			"Model":               c.Param("model"),
-			"Version":             internal.PrintableVersion(),
-		}
-
-		// Render index
-		return c.Render(200, "views/tts", summary)
-	})
-
-	app.GET("/tts", func(c echo.Context) error {
-		modelConfigs := cl.GetAllModelsConfigs()
-		modelsWithoutConfig, _ := services.ListModels(cl, ml, config.NoFilterFn, services.LOOSE_ONLY)
-
-		if len(modelConfigs)+len(modelsWithoutConfig) == 0 {
-			// If no model is available redirect to the index which suggests how to install models
-			return c.Redirect(302, middleware.BaseURL(c))
-		}
-
-		modelThatCanBeUsed := ""
-		title := "LocalAI - Generate audio"
-
-		for _, b := range modelConfigs {
-			if b.HasUsecases(config.FLAG_TTS) {
-				modelThatCanBeUsed = b.Name
-				title = "LocalAI - Generate audio with " + modelThatCanBeUsed
-				break
-			}
-		}
-		summary := map[string]interface{}{
-			"Title":               title,
-			"BaseURL":             middleware.BaseURL(c),
-			"ModelsConfig":        modelConfigs,
-			"ModelsWithoutConfig": modelsWithoutConfig,
-			"Model":               modelThatCanBeUsed,
-			"Version":             internal.PrintableVersion(),
-		}
-
-		// Render index
-		return c.Render(200, "views/tts", summary)
-	})
-
-	app.GET("/video/:model", func(c echo.Context) error {
-		modelConfigs := cl.GetAllModelsConfigs()
-		modelsWithoutConfig, _ := services.ListModels(cl, ml, config.NoFilterFn, services.LOOSE_ONLY)
-
-		summary := map[string]interface{}{
-			"Title":               "LocalAI - Generate videos with " + c.Param("model"),
-			"BaseURL":             middleware.BaseURL(c),
-			"ModelsConfig":        modelConfigs,
-			"ModelsWithoutConfig": modelsWithoutConfig,
-			"Model":               c.Param("model"),
-			"Version":             internal.PrintableVersion(),
-		}
-
-		// Render index
-		return c.Render(200, "views/video", summary)
-	})
-
-	app.GET("/video", func(c echo.Context) error {
-		modelConfigs := cl.GetAllModelsConfigs()
-		modelsWithoutConfig, _ := services.ListModels(cl, ml, config.NoFilterFn, services.LOOSE_ONLY)
-
-		if len(modelConfigs)+len(modelsWithoutConfig) == 0 {
-			// If no model is available redirect to the index which suggests how to install models
-			return c.Redirect(302, middleware.BaseURL(c))
-		}
-
-		modelThatCanBeUsed := ""
-		title := "LocalAI - Generate videos"
-
-		for _, b := range modelConfigs {
-			if b.HasUsecases(config.FLAG_VIDEO) {
-				modelThatCanBeUsed = b.Name
-				title = "LocalAI - Generate videos with " + modelThatCanBeUsed
-				break
-			}
-		}
-
-		summary := map[string]interface{}{
-			"Title":               title,
-			"BaseURL":             middleware.BaseURL(c),
-			"ModelsConfig":        modelConfigs,
-			"ModelsWithoutConfig": modelsWithoutConfig,
-			"Model":               modelThatCanBeUsed,
-			"Version":             internal.PrintableVersion(),
-		}
-
-		// Render index
-		return c.Render(200, "views/video", summary)
-	})
+	app.GET("/tts", localai.WelcomeEndpoint(appConfig, cl, ml, processingOps))

 	// Traces UI
 	app.GET("/traces", func(c echo.Context) error {
--- a/core/http/static/spa-home.js
+++ b/core/http/static/spa-home.js
@@ -0,0 +1,411 @@
+/**
+ * SPA Home View JavaScript
+ * Contains Alpine.js components and functions for the home view
+ */
+
+// Home input form component
+function homeInputForm() {
+  return {
+    selectedModel: '',
+    inputValue: '',
+    shiftPressed: false,
+    fileName: '',
+    imageFiles: [],
+    audioFiles: [],
+    textFiles: [],
+    attachedFiles: [],
+    mcpMode: false,
+    mcpAvailable: false,
+    mcpModels: {},
+    currentPlaceholder: 'Send a message...',
+    placeholderIndex: 0,
+    charIndex: 0,
+    isTyping: false,
+    typingTimeout: null,
+    displayTimeout: null,
+    placeholderMessages: [
+      'What is Nuclear fusion?',
+      'How does a combustion engine work?',
+      'Explain quantum computing',
+      'What causes climate change?',
+      'How do neural networks learn?',
+      'What is the theory of relativity?',
+      'How does photosynthesis work?',
+      'Explain the water cycle',
+      'What is machine learning?',
+      'How do black holes form?',
+      'What is DNA and how does it work?',
+      'Explain the greenhouse effect',
+      'How does the immune system work?',
+      'What is artificial intelligence?',
+      'How do solar panels generate electricity?',
+      'Explain the process of evolution',
+      'What is the difference between weather and climate?',
+      'How does the human brain process information?',
+      'What is the structure of an atom?',
+      'How do vaccines work?',
+      'Explain the concept of entropy',
+      'What is the speed of light?',
+      'How does gravity work?',
+      'What is the difference between mass and weight?'
+    ],
+
+    init() {
+      window.currentPlaceholderText = this.currentPlaceholder;
+      this.startTypingAnimation();
+      // Build MCP models map from data attributes
+      this.buildMCPModelsMap();
+      // Select first model by default
+      this.$nextTick(() => {
+        const select = this.$el.querySelector('select');
+        if (select && select.options.length > 1) {
+          const firstModelOption = select.options[1];
+          if (firstModelOption && firstModelOption.value) {
+            this.selectedModel = firstModelOption.value;
+            this.checkMCPAvailability();
+          }
+        }
+      });
+      // Watch for changes to selectedModel to update MCP availability
+      this.$watch('selectedModel', () => {
+        this.checkMCPAvailability();
+      });
+    },
+
+    buildMCPModelsMap() {
+      const select = this.$el.querySelector('select');
+      if (!select) return;
+      this.mcpModels = {};
+      for (let i = 0; i < select.options.length; i++) {
+        const option = select.options[i];
+        if (option.value) {
+          const hasMcpAttr = option.getAttribute('data-has-mcp');
+          this.mcpModels[option.value] = hasMcpAttr === 'true';
+        }
+      }
+    },
+
+    checkMCPAvailability() {
+      if (!this.selectedModel) {
+        this.mcpAvailable = false;
+        this.mcpMode = false;
+        return;
+      }
+      const hasMCP = this.mcpModels[this.selectedModel] === true;
+      this.mcpAvailable = hasMCP;
+      if (!hasMCP) {
+        this.mcpMode = false;
+      }
+    },
+
+    startTypingAnimation() {
+      if (this.isTyping) return;
+      this.typeNextPlaceholder();
+    },
+
+    typeNextPlaceholder() {
+      if (this.isTyping) return;
+      this.isTyping = true;
+      this.charIndex = 0;
+      const message = this.placeholderMessages[this.placeholderIndex];
+      this.currentPlaceholder = '';
+      window.currentPlaceholderText = '';
+
+      const typeChar = () => {
+        if (this.charIndex < message.length) {
+          this.currentPlaceholder = message.substring(0, this.charIndex + 1);
+          window.currentPlaceholderText = this.currentPlaceholder;
+          this.charIndex++;
+          this.typingTimeout = setTimeout(typeChar, 30);
+        } else {
+          this.isTyping = false;
+          window.currentPlaceholderText = this.currentPlaceholder;
+          this.displayTimeout = setTimeout(() => {
+            this.placeholderIndex = (this.placeholderIndex + 1) % this.placeholderMessages.length;
+            this.typeNextPlaceholder();
+          }, 2000);
+        }
+      };
+
+      typeChar();
+    },
+
+    pauseTyping() {
+      if (this.typingTimeout) {
+        clearTimeout(this.typingTimeout);
+        this.typingTimeout = null;
+      }
+      if (this.displayTimeout) {
+        clearTimeout(this.displayTimeout);
+        this.displayTimeout = null;
+      }
+      this.isTyping = false;
+    },
+
+    resumeTyping() {
+      if (!this.inputValue.trim() && !this.isTyping) {
+        this.startTypingAnimation();
+      }
+    },
+
+    handleFocus() {
+      if (this.isTyping && this.placeholderIndex < this.placeholderMessages.length) {
+        const fullMessage = this.placeholderMessages[this.placeholderIndex];
+        this.currentPlaceholder = fullMessage;
+        window.currentPlaceholderText = fullMessage;
+      }
+      this.pauseTyping();
+    },
+
+    handleBlur() {
+      if (!this.inputValue.trim()) {
+        this.resumeTyping();
+      }
+    },
+
+    handleInput() {
+      if (this.inputValue.trim()) {
+        this.pauseTyping();
+      } else {
+        this.resumeTyping();
+      }
+    },
+
+    handleFileSelection(files, fileType) {
+      Array.from(files).forEach(file => {
+        const exists = this.attachedFiles.some(f => f.name === file.name && f.type === fileType);
+        if (!exists) {
+          this.attachedFiles.push({ name: file.name, type: fileType });
+        }
+      });
+    },
+
+    removeAttachedFile(fileType, fileName) {
+      const index = this.attachedFiles.findIndex(f => f.name === fileName && f.type === fileType);
+      if (index !== -1) {
+        this.attachedFiles.splice(index, 1);
+      }
+      if (fileType === 'image') {
+        this.imageFiles = this.imageFiles.filter(f => f.name !== fileName);
+      } else if (fileType === 'audio') {
+        this.audioFiles = this.audioFiles.filter(f => f.name !== fileName);
+      } else if (fileType === 'file') {
+        this.textFiles = this.textFiles.filter(f => f.name !== fileName);
+      }
+    }
+  };
+}
+
+// Start chat function for SPA - navigates to chat view instead of full page redirect
+function startChatSPA(event) {
+  if (event) {
+    event.preventDefault();
+  }
+
+  const form = event ? event.target.closest('form') : document.querySelector('form');
+  if (!form) return;
+
+  const alpineComponent = form.closest('[x-data]');
+  const select = alpineComponent ? alpineComponent.querySelector('select') : null;
+  const textarea = form.querySelector('textarea');
+
+  const selectedModel = select ? select.value : '';
+  let message = textarea ? textarea.value : '';
+
+  if (!message.trim() && window.currentPlaceholderText) {
+    message = window.currentPlaceholderText;
+  }
+
+  if (!selectedModel || !message.trim()) {
+    return;
+  }
+
+  // Get MCP mode from checkbox
+  let mcpMode = false;
+  const mcpToggle = document.getElementById('spa_home_mcp_toggle');
+  if (mcpToggle && mcpToggle.checked) {
+    mcpMode = true;
+  }
+
+  // Store message and files in localStorage for chat view to pick up
+  const chatData = {
+    message: message,
+    imageFiles: [],
+    audioFiles: [],
+    textFiles: [],
+    mcpMode: mcpMode
+  };
+
+  // Convert files to base64 for storage
+  const imageInput = document.getElementById('spa_home_input_image');
+  const audioInput = document.getElementById('spa_home_input_audio');
+  const fileInput = document.getElementById('spa_home_input_file');
+
+  const filePromises = [
+    ...Array.from(imageInput?.files || []).map(file =>
+      new Promise(resolve => {
+        const reader = new FileReader();
+        reader.onload = e => resolve({ name: file.name, data: e.target.result, type: file.type });
+        reader.readAsDataURL(file);
+      })
+    ),
+    ...Array.from(audioInput?.files || []).map(file =>
+      new Promise(resolve => {
+        const reader = new FileReader();
+        reader.onload = e => resolve({ name: file.name, data: e.target.result, type: file.type });
+        reader.readAsDataURL(file);
+      })
+    ),
+    ...Array.from(fileInput?.files || []).map(file =>
+      new Promise(resolve => {
+        const reader = new FileReader();
+        reader.onload = e => resolve({ name: file.name, data: e.target.result, type: file.type });
+        reader.readAsText(file);
+      })
+    )
+  ];
+
+  const navigateToChat = () => {
+    // Store in localStorage
+    localStorage.setItem('localai_index_chat_data', JSON.stringify(chatData));
+
+    // Use SPA router to navigate to chat
+    if (window.Alpine && Alpine.store('router')) {
+      Alpine.store('router').navigate('chat', { model: selectedModel });
+    } else {
+      // Fallback to full page redirect if router not available
+      window.location.href = `/chat/${selectedModel}`;
+    }
+  };
+
+  if (filePromises.length > 0) {
+    Promise.all(filePromises).then(files => {
+      files.forEach(file => {
+        if (file.type.startsWith('image/')) {
+          chatData.imageFiles.push(file);
+        } else if (file.type.startsWith('audio/')) {
+          chatData.audioFiles.push(file);
+        } else {
+          chatData.textFiles.push(file);
+        }
+      });
+      navigateToChat();
+    }).catch(err => {
+      console.error('Error processing files:', err);
+      navigateToChat();
+    });
+  } else {
+    navigateToChat();
+  }
+}
+
+// Resource Monitor component (GPU if available, otherwise RAM)
+function resourceMonitor() {
+  return {
+    resourceData: null,
+    pollInterval: null,
+
+    async fetchResourceData() {
+      try {
+        const response = await fetch('/api/resources');
+        if (response.ok) {
+          this.resourceData = await response.json();
+        }
+      } catch (error) {
+        console.error('Error fetching resource data:', error);
+      }
+    },
+
+    startPolling() {
+      this.fetchResourceData();
+      this.pollInterval = setInterval(() => this.fetchResourceData(), 5000);
+    },
+
+    stopPolling() {
+      if (this.pollInterval) {
+        clearInterval(this.pollInterval);
+      }
+    }
+  };
+}
+
+// Stop individual model
+async function stopModel(modelName) {
+  if (!confirm(`Are you sure you want to stop "${modelName}"?`)) {
+    return;
+  }
+
+  try {
+    const response = await fetch('/backend/shutdown', {
+      method: 'POST',
+      headers: {
+        'Content-Type': 'application/json',
+      },
+      body: JSON.stringify({ model: modelName })
+    });
+
+    if (response.ok) {
+      setTimeout(() => {
+        window.location.reload();
+      }, 500);
+    } else {
+      alert('Failed to stop model');
+    }
+  } catch (error) {
+    console.error('Error stopping model:', error);
+    alert('Failed to stop model');
+  }
+}
+
+// Stop all loaded models
+async function stopAllModels(component) {
+  // Get loaded models from DOM
+  const loadedModelElements = document.querySelectorAll('[data-loaded-model]');
+  const loadedModelNames = Array.from(loadedModelElements).map(el => {
+    const span = el.querySelector('span.truncate');
+    return span ? span.textContent.trim() : '';
+  }).filter(name => name.length > 0);
+
+  if (loadedModelNames.length === 0) {
+    return;
+  }
+
+  if (!confirm(`Are you sure you want to stop all ${loadedModelNames.length} loaded model(s)?`)) {
+    return;
+  }
+
+  if (component) {
+    component.stoppingAll = true;
+  }
+
+  try {
+    const stopPromises = loadedModelNames.map(modelName =>
+      fetch('/backend/shutdown', {
+        method: 'POST',
+        headers: {
+          'Content-Type': 'application/json',
+        },
+        body: JSON.stringify({ model: modelName })
+      })
+    );
+
+    await Promise.all(stopPromises);
+
+    setTimeout(() => {
+      window.location.reload();
+    }, 1000);
+  } catch (error) {
+    console.error('Error stopping models:', error);
+    alert('Failed to stop some models');
+    if (component) {
+      component.stoppingAll = false;
+    }
+  }
+}
+
+// Make functions available globally
+window.homeInputForm = homeInputForm;
+window.startChatSPA = startChatSPA;
+window.resourceMonitor = resourceMonitor;
+window.stopModel = stopModel;
+window.stopAllModels = stopAllModels;
--- a/core/http/static/spa-router.js
+++ b/core/http/static/spa-router.js
@@ -0,0 +1,148 @@
+/**
+ * LocalAI SPA Router
+ * Client-side routing for the single-page application
+ */
+
+// Define routes and their corresponding view IDs
+const SPA_ROUTES = {
+  'home': { title: 'LocalAI', viewId: 'view-home', paths: ['/', ''] },
+  'chat': { title: 'LocalAI - Chat', viewId: 'view-chat', paths: ['/chat'] },
+  'text2image': { title: 'LocalAI - Images', viewId: 'view-text2image', paths: ['/text2image'] },
+  'tts': { title: 'LocalAI - TTS', viewId: 'view-tts', paths: ['/tts'] },
+  'talk': { title: 'LocalAI - Talk', viewId: 'view-talk', paths: ['/talk'] },
+  'manage': { title: 'LocalAI - System', viewId: 'view-manage', paths: ['/manage'] },
+  'browse': { title: 'LocalAI - Model Gallery', viewId: 'view-browse', paths: ['/browse'] }
+};
+
+// Parse URL path to determine route
+function parseUrlPath(pathname) {
+  // Remove trailing slash
+  pathname = pathname.replace(/\/$/, '') || '/';
+  
+  // Check for hash-based routes first (for SPA navigation)
+  const hash = window.location.hash.slice(1);
+  if (hash) {
+    const hashParts = hash.split('/');
+    const route = hashParts[0];
+    const model = hashParts[1] || null;
+    if (SPA_ROUTES[route]) {
+      return { route, params: model ? { model } : {} };
+    }
+  }
+  
+  // Check path-based routes
+  for (const [route, config] of Object.entries(SPA_ROUTES)) {
+    for (const path of config.paths) {
+      if (pathname === path) {
+        return { route, params: {} };
+      }
+      // Check for parameterized routes like /chat/:model
+      if (pathname.startsWith(path + '/')) {
+        const param = pathname.slice(path.length + 1);
+        if (param) {
+          return { route, params: { model: param } };
+        }
+      }
+    }
+  }
+  
+  // Default to home
+  return { route: 'home', params: {} };
+}
+
+// Initialize the router store for Alpine.js
+document.addEventListener('alpine:init', () => {
+  // Parse initial route from URL
+  const initialRoute = parseUrlPath(window.location.pathname);
+  
+  Alpine.store('router', {
+    currentRoute: initialRoute.route,
+    routeParams: initialRoute.params,
+    previousRoute: null,
+    
+    /**
+     * Navigate to a route
+     * @param {string} route - The route name to navigate to
+     * @param {Object} params - Optional parameters for the route
+     */
+    navigate(route, params = {}) {
+      if (!SPA_ROUTES[route]) {
+        console.warn(`Unknown route: ${route}`);
+        return;
+      }
+      
+      this.previousRoute = this.currentRoute;
+      this.currentRoute = route;
+      this.routeParams = params;
+      
+      // Update document title
+      document.title = SPA_ROUTES[route].title;
+      
+      // Update URL without page reload using history API
+      const url = route === 'home' ? '/' : `/#${route}`;
+      if (params.model) {
+        window.history.pushState({ route, params }, '', `/#${route}/${params.model}`);
+      } else {
+        window.history.pushState({ route, params }, '', url);
+      }
+      
+      // Scroll to top on navigation
+      window.scrollTo(0, 0);
+      
+      // Emit custom event for route change listeners
+      window.dispatchEvent(new CustomEvent('spa:navigate', { 
+        detail: { route, params, previousRoute: this.previousRoute } 
+      }));
+    },
+    
+    /**
+     * Check if the current route matches
+     * @param {string} route - The route to check
+     * @returns {boolean}
+     */
+    isRoute(route) {
+      return this.currentRoute === route;
+    },
+    
+    /**
+     * Navigate to chat with a specific model
+     * @param {string} model - The model name
+     */
+    navigateToChat(model) {
+      this.navigate('chat', { model });
+    },
+    
+    /**
+     * Navigate to text2image with a specific model
+     * @param {string} model - The model name
+     */
+    navigateToText2Image(model) {
+      this.navigate('text2image', { model });
+    },
+    
+    /**
+     * Navigate to TTS with a specific model
+     * @param {string} model - The model name
+     */
+    navigateToTTS(model) {
+      this.navigate('tts', { model });
+    }
+  });
+});
+
+// Handle browser back/forward buttons
+window.addEventListener('popstate', (event) => {
+  if (event.state && event.state.route) {
+    Alpine.store('router').currentRoute = event.state.route;
+    Alpine.store('router').routeParams = event.state.params || {};
+  } else {
+    // Parse URL for route
+    const parsed = parseUrlPath(window.location.pathname);
+    Alpine.store('router').currentRoute = parsed.route;
+    Alpine.store('router').routeParams = parsed.params;
+  }
+});
+
+// Export for use in other scripts
+window.SPA_ROUTES = SPA_ROUTES;
+window.parseUrlPath = parseUrlPath;
--- a/core/http/static/video.js
+++ b/core/http/static/video.js
@@ -1,376 +0,0 @@
-// Helper function to convert file to base64
-function fileToBase64(file) {
-  return new Promise((resolve, reject) => {
-    const reader = new FileReader();
-    reader.onload = () => {
-      // Remove data:image/...;base64, prefix if present
-      const base64 = reader.result.split(',')[1] || reader.result;
-      resolve(base64);
-    };
-    reader.onerror = reject;
-    reader.readAsDataURL(file);
-  });
-}
-
-function genVideo(event) {
-  event.preventDefault();
-  promptVideo();
-}
-
-async function promptVideo() {
-  const loader = document.getElementById("loader");
-  const input = document.getElementById("input");
-  const generateBtn = document.getElementById("generate-btn");
-  const resultDiv = document.getElementById("result");
-  const resultPlaceholder = document.getElementById("result-placeholder");
-
-  // Show loader and disable form
-  loader.classList.remove("hidden");
-  if (resultPlaceholder) {
-    resultPlaceholder.style.display = "none";
-  }
-  input.disabled = true;
-  generateBtn.disabled = true;
-
-  // Store the prompt for later restoration
-  const prompt = input.value.trim();
-  if (!prompt) {
-    alert("Please enter a prompt");
-    loader.classList.add("hidden");
-    if (resultPlaceholder) {
-      resultPlaceholder.style.display = "flex";
-    }
-    input.disabled = false;
-    generateBtn.disabled = false;
-    return;
-  }
-
-  // Collect all form values
-  const model = document.getElementById("video-model").value;
-  const size = document.getElementById("video-size").value;
-  const negativePrompt = document.getElementById("negative-prompt").value.trim();
-  
-  // Parse size into width and height
-  const sizeParts = size.split("x");
-  let width = 512;
-  let height = 512;
-  if (sizeParts.length === 2) {
-    width = parseInt(sizeParts[0]) || 512;
-    height = parseInt(sizeParts[1]) || 512;
-  }
-
-  // Video-specific parameters
-  const secondsInput = document.getElementById("video-seconds").value.trim();
-  const seconds = secondsInput ? secondsInput : undefined;
-  const fpsInput = document.getElementById("video-fps").value.trim();
-  const fps = fpsInput ? parseInt(fpsInput) : 16;
-  const framesInput = document.getElementById("video-frames").value.trim();
-  const numFrames = framesInput ? parseInt(framesInput) : undefined;
-
-  // Advanced parameters
-  const stepInput = document.getElementById("video-steps").value.trim();
-  const step = stepInput ? parseInt(stepInput) : undefined;
-  const seedInput = document.getElementById("video-seed").value.trim();
-  const seed = seedInput ? parseInt(seedInput) : undefined;
-  const cfgScaleInput = document.getElementById("video-cfg-scale").value.trim();
-  const cfgScale = cfgScaleInput ? parseFloat(cfgScaleInput) : undefined;
-
-  // Prepare request body
-  const requestBody = {
-    model: model,
-    prompt: prompt,
-    width: width,
-    height: height,
-    fps: fps,
-  };
-
-  if (negativePrompt) {
-    requestBody.negative_prompt = negativePrompt;
-  }
-
-  if (seconds !== undefined) {
-    requestBody.seconds = seconds;
-  }
-
-  if (numFrames !== undefined) {
-    requestBody.num_frames = numFrames;
-  }
-
-  if (step !== undefined) {
-    requestBody.step = step;
-  }
-
-  if (seed !== undefined) {
-    requestBody.seed = seed;
-  }
-
-  if (cfgScale !== undefined) {
-    requestBody.cfg_scale = cfgScale;
-  }
-
-  // Handle file inputs
-  try {
-    // Start image (for img2video)
-    const startImageInput = document.getElementById("start-image");
-    if (startImageInput.files.length > 0) {
-      const base64 = await fileToBase64(startImageInput.files[0]);
-      requestBody.start_image = base64;
-    }
-
-    // End image
-    const endImageInput = document.getElementById("end-image");
-    if (endImageInput.files.length > 0) {
-      const base64 = await fileToBase64(endImageInput.files[0]);
-      requestBody.end_image = base64;
-    }
-  } catch (error) {
-    console.error("Error processing image files:", error);
-    resultDiv.innerHTML = '<p class="text-xs text-red-500 p-2">Error processing image files: ' + error.message + '</p>';
-    loader.classList.add("hidden");
-    if (resultPlaceholder) {
-      resultPlaceholder.style.display = "none";
-    }
-    input.disabled = false;
-    generateBtn.disabled = false;
-    return;
-  }
-
-  // Make API request to LocalAI endpoint
-  try {
-    const response = await fetch("video", {
-      method: "POST",
-      headers: {
-        "Content-Type": "application/json",
-      },
-      body: JSON.stringify(requestBody),
-    });
-
-    const json = await response.json();
-
-    if (json.error) {
-      // Display error
-      resultDiv.innerHTML = '<p class="text-xs text-red-500 p-2">Error: ' + json.error.message + '</p>';
-      loader.classList.add("hidden");
-      if (resultPlaceholder) {
-        resultPlaceholder.style.display = "none";
-      }
-      input.disabled = false;
-      generateBtn.disabled = false;
-      return;
-    }
-
-    // Clear result div and hide placeholder
-    resultDiv.innerHTML = '';
-    if (resultPlaceholder) {
-      resultPlaceholder.style.display = "none";
-    }
-
-    // Display generated video
-    if (json.data && json.data.length > 0) {
-      json.data.forEach((item, index) => {
-        const videoContainer = document.createElement("div");
-        videoContainer.className = "flex flex-col";
-
-        // Create video element
-        const video = document.createElement("video");
-        video.controls = true;
-        video.className = "w-full h-auto rounded-lg";
-        video.preload = "metadata";
-        
-        if (item.url) {
-          video.src = item.url;
-        } else if (item.b64_json) {
-          video.src = "data:video/mp4;base64," + item.b64_json;
-        } else {
-          return; // Skip invalid items
-        }
-        
-        videoContainer.appendChild(video);
-
-        // Create caption container
-        const captionDiv = document.createElement("div");
-        captionDiv.className = "mt-2 p-2 bg-[var(--color-bg-secondary)] rounded-lg text-xs";
-
-        // Prompt caption
-        const promptCaption = document.createElement("p");
-        promptCaption.className = "text-[var(--color-text-primary)] mb-1.5 break-words";
-        promptCaption.innerHTML = '<strong>Prompt:</strong> ' + escapeHtml(prompt);
-        captionDiv.appendChild(promptCaption);
-
-        // Negative prompt if provided
-        if (negativePrompt) {
-          const negativeCaption = document.createElement("p");
-          negativeCaption.className = "text-[var(--color-text-secondary)] mb-1.5 break-words";
-          negativeCaption.innerHTML = '<strong>Negative Prompt:</strong> ' + escapeHtml(negativePrompt);
-          captionDiv.appendChild(negativeCaption);
-        }
-
-        // Generation details
-        const detailsDiv = document.createElement("div");
-        detailsDiv.className = "flex flex-wrap gap-3 text-[10px] text-[var(--color-text-secondary)] mt-1.5";
-        detailsDiv.innerHTML = `
-          <span><strong>Size:</strong> ${width}x${height}</span>
-          ${fps ? `<span><strong>FPS:</strong> ${fps}</span>` : ''}
-          ${numFrames !== undefined ? `<span><strong>Frames:</strong> ${numFrames}</span>` : ''}
-          ${seconds !== undefined ? `<span><strong>Duration:</strong> ${seconds}s</span>` : ''}
-          ${step !== undefined ? `<span><strong>Steps:</strong> ${step}</span>` : ''}
-          ${seed !== undefined ? `<span><strong>Seed:</strong> ${seed}</span>` : ''}
-          ${cfgScale !== undefined ? `<span><strong>CFG Scale:</strong> ${cfgScale}</span>` : ''}
-        `;
-        captionDiv.appendChild(detailsDiv);
-
-        // Button container
-        const buttonContainer = document.createElement("div");
-        buttonContainer.className = "mt-1.5 flex gap-2";
-
-        // Copy prompt button
-        const copyBtn = document.createElement("button");
-        copyBtn.className = "px-2 py-0.5 text-[10px] bg-[var(--color-primary)] text-white rounded hover:opacity-80";
-        copyBtn.innerHTML = '<i class="fas fa-copy mr-1"></i>Copy Prompt';
-        copyBtn.onclick = () => {
-          navigator.clipboard.writeText(prompt).then(() => {
-            copyBtn.innerHTML = '<i class="fas fa-check mr-1"></i>Copied!';
-            setTimeout(() => {
-              copyBtn.innerHTML = '<i class="fas fa-copy mr-1"></i>Copy Prompt';
-            }, 2000);
-          });
-        };
-        buttonContainer.appendChild(copyBtn);
-
-        // Download video button
-        const downloadBtn = document.createElement("button");
-        downloadBtn.className = "px-2 py-0.5 text-[10px] bg-[var(--color-primary)] text-white rounded hover:opacity-80";
-        downloadBtn.innerHTML = '<i class="fas fa-download mr-1"></i>Download Video';
-        downloadBtn.onclick = () => {
-          downloadVideo(item, downloadBtn);
-        };
-        buttonContainer.appendChild(downloadBtn);
-
-        captionDiv.appendChild(buttonContainer);
-
-        videoContainer.appendChild(captionDiv);
-        resultDiv.appendChild(videoContainer);
-      });
-      // Hide placeholder when videos are displayed
-      if (resultPlaceholder) {
-        resultPlaceholder.style.display = "none";
-      }
-    } else {
-      resultDiv.innerHTML = '<p class="text-xs text-[var(--color-text-secondary)] p-2">No videos were generated.</p>';
-      if (resultPlaceholder) {
-        resultPlaceholder.style.display = "none";
-      }
-    }
-
-  } catch (error) {
-    console.error("Error generating video:", error);
-    resultDiv.innerHTML = '<p class="text-xs text-red-500 p-2">Error: ' + error.message + '</p>';
-    if (resultPlaceholder) {
-      resultPlaceholder.style.display = "none";
-    }
-  } finally {
-    // Hide loader and re-enable form
-    loader.classList.add("hidden");
-    input.disabled = false;
-    generateBtn.disabled = false;
-    input.focus();
-  }
-}
-
-// Helper function to escape HTML
-function escapeHtml(text) {
-  const div = document.createElement("div");
-  div.textContent = text;
-  return div.innerHTML;
-}
-
-// Helper function to download video
-function downloadVideo(item, button) {
-  try {
-    let videoUrl;
-    let filename = "generated-video.mp4";
-
-    if (item.url) {
-      // If we have a URL, use it directly
-      videoUrl = item.url;
-      // Extract filename from URL if possible
-      const urlParts = item.url.split("/");
-      if (urlParts.length > 0) {
-        const lastPart = urlParts[urlParts.length - 1];
-        if (lastPart && lastPart.includes(".")) {
-          filename = lastPart;
-        }
-      }
-    } else if (item.b64_json) {
-      // Convert base64 to blob
-      const byteCharacters = atob(item.b64_json);
-      const byteNumbers = new Array(byteCharacters.length);
-      for (let i = 0; i < byteCharacters.length; i++) {
-        byteNumbers[i] = byteCharacters.charCodeAt(i);
-      }
-      const byteArray = new Uint8Array(byteNumbers);
-      const blob = new Blob([byteArray], { type: "video/mp4" });
-      videoUrl = URL.createObjectURL(blob);
-    } else {
-      console.error("No video data available for download");
-      return;
-    }
-
-    // Create a temporary anchor element to trigger download
-    const link = document.createElement("a");
-    link.href = videoUrl;
-    link.download = filename;
-    link.style.display = "none";
-    document.body.appendChild(link);
-    link.click();
-    document.body.removeChild(link);
-
-    // Clean up object URL if we created one
-    if (item.b64_json && videoUrl.startsWith("blob:")) {
-      setTimeout(() => URL.revokeObjectURL(videoUrl), 100);
-    }
-
-    // Show feedback
-    const originalHTML = button.innerHTML;
-    button.innerHTML = '<i class="fas fa-check mr-1"></i>Downloaded!';
-    setTimeout(() => {
-      button.innerHTML = originalHTML;
-    }, 2000);
-  } catch (error) {
-    console.error("Error downloading video:", error);
-    button.innerHTML = '<i class="fas fa-exclamation-triangle mr-1"></i>Error';
-    setTimeout(() => {
-      button.innerHTML = '<i class="fas fa-download mr-1"></i>Download Video';
-    }, 2000);
-  }
-}
-
-// Initialize
-document.addEventListener("DOMContentLoaded", function() {
-  const input = document.getElementById("input");
-  const form = document.getElementById("genvideo");
-
-  if (input) {
-    input.focus();
-  }
-
-  if (form) {
-    form.addEventListener("submit", genVideo);
-  }
-
-  // Handle Enter key press in the prompt input (but allow Shift+Enter for new lines)
-  if (input) {
-    input.addEventListener("keydown", function(event) {
-      if (event.key === "Enter" && !event.shiftKey) {
-        event.preventDefault();
-        genVideo(event);
-      }
-    });
-  }
-
-  // Hide loader initially
-  const loader = document.getElementById("loader");
-  if (loader) {
-    loader.classList.add("hidden");
-  }
-});
--- a/core/http/views/manage.html
+++ b/core/http/views/manage.html
@@ -315,7 +315,7 @@
                                    </a>
                                    {{ end }}
                                    {{ if eq . "FLAG_IMAGE" }}
-                                    <a href="image/{{$backendCfg.Name}}" class="inline-flex items-center px-1.5 py-0.5 rounded text-[10px] font-medium bg-[var(--color-success)]/10 text-green-300 hover:bg-[var(--color-success)]/20 transition-colors" title="Image">
+                                    <a href="text2image/{{$backendCfg.Name}}" class="inline-flex items-center px-1.5 py-0.5 rounded text-[10px] font-medium bg-[var(--color-success)]/10 text-green-300 hover:bg-[var(--color-success)]/20 transition-colors" title="Image">
                                        <i class="fas fa-image text-[8px] mr-1"></i>Image
                                    </a>
                                    {{ end }}
--- a/core/http/views/partials/navbar.html
+++ b/core/http/views/partials/navbar.html
@@ -25,12 +25,9 @@
                <a href="chat/" class="text-[var(--color-text-secondary)] hover:text-[var(--color-text-primary)] px-2 py-2 rounded-lg transition duration-300 ease-in-out hover:bg-[var(--color-bg-secondary)] flex items-center group text-sm">
                    <i class="fa-solid fa-comments text-[var(--color-primary)] mr-1.5 text-sm group-hover:scale-110 transition-transform"></i>Chat
                </a>
-                <a href="image/" class="text-[var(--color-text-secondary)] hover:text-[var(--color-text-primary)] px-2 py-2 rounded-lg transition duration-300 ease-in-out hover:bg-[var(--color-bg-secondary)] flex items-center group text-sm">
+                <a href="text2image/" class="text-[var(--color-text-secondary)] hover:text-[var(--color-text-primary)] px-2 py-2 rounded-lg transition duration-300 ease-in-out hover:bg-[var(--color-bg-secondary)] flex items-center group text-sm">
                    <i class="fas fa-image text-[var(--color-primary)] mr-1.5 text-sm group-hover:scale-110 transition-transform"></i>Images
                </a>
-                <a href="video/" class="text-[var(--color-text-secondary)] hover:text-[var(--color-text-primary)] px-2 py-2 rounded-lg transition duration-300 ease-in-out hover:bg-[var(--color-bg-secondary)] flex items-center group text-sm">
-                    <i class="fas fa-video text-[var(--color-primary)] mr-1.5 text-sm group-hover:scale-110 transition-transform"></i>Video
-                </a>
                <a href="tts/" class="text-[var(--color-text-secondary)] hover:text-[var(--color-text-primary)] px-2 py-2 rounded-lg transition duration-300 ease-in-out hover:bg-[var(--color-bg-secondary)] flex items-center group text-sm">
                    <i class="fa-solid fa-music text-[var(--color-primary)] mr-1.5 text-sm group-hover:scale-110 transition-transform"></i>TTS
                </a>
@@ -88,12 +85,9 @@
                <a href="chat/" class="block text-[var(--color-text-secondary)] hover:text-[var(--color-text-primary)] hover:bg-[var(--color-bg-secondary)] px-3 py-2 rounded-lg transition duration-300 ease-in-out flex items-center text-sm">
                    <i class="fa-solid fa-comments text-[var(--color-primary)] mr-3 w-5 text-center text-sm"></i>Chat
                </a>
-                <a href="image/" class="block text-[var(--color-text-secondary)] hover:text-[var(--color-text-primary)] hover:bg-[var(--color-bg-secondary)] px-3 py-2 rounded-lg transition duration-300 ease-in-out flex items-center text-sm">
+                <a href="text2image/" class="block text-[var(--color-text-secondary)] hover:text-[var(--color-text-primary)] hover:bg-[var(--color-bg-secondary)] px-3 py-2 rounded-lg transition duration-300 ease-in-out flex items-center text-sm">
                    <i class="fas fa-image text-[var(--color-primary)] mr-3 w-5 text-center text-sm"></i>Images
                </a>
-                <a href="video/" class="block text-[var(--color-text-secondary)] hover:text-[var(--color-text-primary)] hover:bg-[var(--color-bg-secondary)] px-3 py-2 rounded-lg transition duration-300 ease-in-out flex items-center text-sm">
-                    <i class="fas fa-video text-[var(--color-primary)] mr-3 w-5 text-center text-sm"></i>Video
-                </a>
                <a href="tts/" class="block text-[var(--color-text-secondary)] hover:text-[var(--color-text-primary)] hover:bg-[var(--color-bg-secondary)] px-3 py-2 rounded-lg transition duration-300 ease-in-out flex items-center text-sm">
                    <i class="fa-solid fa-music text-[var(--color-primary)] mr-3 w-5 text-center text-sm"></i>TTS
                </a>
--- a/core/http/views/partials/spa_navbar.html
+++ b/core/http/views/partials/spa_navbar.html
@@ -0,0 +1,154 @@
+<nav class="bg-[var(--color-bg-primary)] shadow-2xl border-b border-[var(--color-bg-secondary)]">
+    <div class="container mx-auto px-4 py-2">
+        <div class="flex items-center justify-between">
+            <div class="flex items-center">
+                <!-- Logo Image -->
+                <a href="#" @click.prevent="$store.router.navigate('home')" class="flex items-center group">
+                    <img src="static/logo_horizontal.png" 
+                         alt="LocalAI Logo" 
+                         class="h-10 mr-3 brightness-110 transition-all duration-300 group-hover:brightness-125 group-hover:drop-shadow-[0_0_8px_var(--color-primary-border)]">
+                </a>
+            </div>
+            
+            <!-- Menu button for small screens -->
+            <div class="lg:hidden">
+                <button @click="mobileMenuOpen = !mobileMenuOpen" class="text-[var(--color-text-secondary)] hover:text-[var(--color-primary)] focus:outline-none p-2 rounded-lg transition duration-300 ease-in-out hover:bg-[var(--color-bg-secondary)]">
+                    <i class="fas fa-bars fa-lg"></i>
+                </button>
+            </div>
+            
+            <!-- Navigation links -->
+            <div class="hidden lg:flex lg:items-center lg:justify-end lg:space-x-1">
+                <a href="#" @click.prevent="$store.router.navigate('home')" 
+                   :class="$store.router.currentRoute === 'home' ? 'text-[var(--color-primary)] bg-[var(--color-bg-secondary)]' : 'text-[var(--color-text-secondary)]'"
+                   class="hover:text-[var(--color-text-primary)] px-2 py-2 rounded-lg transition duration-300 ease-in-out hover:bg-[var(--color-bg-secondary)] flex items-center group text-sm">
+                    <i class="fas fa-home text-[var(--color-primary)] mr-1.5 text-sm group-hover:scale-110 transition-transform"></i>Home
+                </a>
+                <a href="#" @click.prevent="$store.router.navigate('chat')"
+                   :class="$store.router.currentRoute === 'chat' ? 'text-[var(--color-primary)] bg-[var(--color-bg-secondary)]' : 'text-[var(--color-text-secondary)]'"
+                   class="hover:text-[var(--color-text-primary)] px-2 py-2 rounded-lg transition duration-300 ease-in-out hover:bg-[var(--color-bg-secondary)] flex items-center group text-sm">
+                    <i class="fa-solid fa-comments text-[var(--color-primary)] mr-1.5 text-sm group-hover:scale-110 transition-transform"></i>Chat
+                </a>
+                <a href="#" @click.prevent="$store.router.navigate('text2image')"
+                   :class="$store.router.currentRoute === 'text2image' ? 'text-[var(--color-primary)] bg-[var(--color-bg-secondary)]' : 'text-[var(--color-text-secondary)]'"
+                   class="hover:text-[var(--color-text-primary)] px-2 py-2 rounded-lg transition duration-300 ease-in-out hover:bg-[var(--color-bg-secondary)] flex items-center group text-sm">
+                    <i class="fas fa-image text-[var(--color-primary)] mr-1.5 text-sm group-hover:scale-110 transition-transform"></i>Images
+                </a>
+                <a href="#" @click.prevent="$store.router.navigate('tts')"
+                   :class="$store.router.currentRoute === 'tts' ? 'text-[var(--color-primary)] bg-[var(--color-bg-secondary)]' : 'text-[var(--color-text-secondary)]'"
+                   class="hover:text-[var(--color-text-primary)] px-2 py-2 rounded-lg transition duration-300 ease-in-out hover:bg-[var(--color-bg-secondary)] flex items-center group text-sm">
+                    <i class="fa-solid fa-music text-[var(--color-primary)] mr-1.5 text-sm group-hover:scale-110 transition-transform"></i>TTS
+                </a>
+                <a href="#" @click.prevent="$store.router.navigate('talk')"
+                   :class="$store.router.currentRoute === 'talk' ? 'text-[var(--color-primary)] bg-[var(--color-bg-secondary)]' : 'text-[var(--color-text-secondary)]'"
+                   class="hover:text-[var(--color-text-primary)] px-2 py-2 rounded-lg transition duration-300 ease-in-out hover:bg-[var(--color-bg-secondary)] flex items-center group text-sm">
+                    <i class="fa-solid fa-phone text-[var(--color-primary)] mr-1.5 text-sm group-hover:scale-110 transition-transform"></i>Talk
+                </a>
+                <a href="agent-jobs" class="text-[var(--color-text-secondary)] hover:text-[var(--color-text-primary)] px-2 py-2 rounded-lg transition duration-300 ease-in-out hover:bg-[var(--color-bg-secondary)] flex items-center group text-sm">
+                    <i class="fas fa-tasks text-[var(--color-primary)] mr-1.5 text-sm group-hover:scale-110 transition-transform"></i>Agent Jobs
+                </a>
+                <a href="traces/" class="text-[var(--color-text-secondary)] hover:text-[var(--color-text-primary)] px-2 py-2 rounded-lg transition duration-300 ease-in-out hover:bg-[var(--color-bg-secondary)] flex items-center group text-sm">
+                    <i class="fas fa-chart-line text-[var(--color-primary)] mr-1.5 text-sm group-hover:scale-110 transition-transform"></i>Traces
+                </a>
+                <a href="swagger/index.html" class="text-[var(--color-text-secondary)] hover:text-[var(--color-text-primary)] px-2 py-2 rounded-lg transition duration-300 ease-in-out hover:bg-[var(--color-bg-secondary)] flex items-center group text-sm">
+                    <i class="fas fa-code text-[var(--color-primary)] mr-1.5 text-sm group-hover:scale-110 transition-transform"></i>API
+                </a>
+                
+                <!-- System Dropdown -->
+                <div class="relative" @click.away="settingsOpen = false">
+                    <button @click="settingsOpen = !settingsOpen" 
+                            class="text-[var(--color-text-secondary)] hover:text-[var(--color-text-primary)] px-2 py-2 rounded-lg transition duration-300 ease-in-out hover:bg-[var(--color-bg-secondary)] flex items-center group text-sm">
+                        <i class="fas fa-cog text-[var(--color-primary)] mr-1.5 text-sm group-hover:scale-110 transition-transform"></i>Settings
+                        <i class="fas fa-chevron-down ml-1 text-xs transition-transform" :class="settingsOpen ? 'rotate-180' : ''"></i>
+                    </button>
+                    <div x-show="settingsOpen" 
+                         x-transition:enter="transition ease-out duration-200"
+                         x-transition:enter-start="opacity-0 scale-95"
+                         x-transition:enter-end="opacity-100 scale-100"
+                         x-transition:leave="transition ease-in duration-150"
+                         x-transition:leave-start="opacity-100 scale-100"
+                         x-transition:leave-end="opacity-0 scale-95"
+                         class="absolute top-full right-0 mt-1 w-48 bg-[var(--color-bg-secondary)] border border-[var(--color-primary-border)]/20 rounded-lg shadow-lg z-50 py-1">
+                        <a href="#" @click.prevent="$store.router.navigate('browse'); settingsOpen = false" class="block text-[var(--color-text-secondary)] hover:text-[var(--color-text-primary)] hover:bg-[var(--color-bg-primary)] px-3 py-2 text-sm transition-colors flex items-center">
+                            <i class="fas fa-brain text-[var(--color-primary)] mr-2 text-xs"></i>Models
+                        </a>
+                        <a href="browse/backends" class="block text-[var(--color-text-secondary)] hover:text-[var(--color-text-primary)] hover:bg-[var(--color-bg-primary)] px-3 py-2 text-sm transition-colors flex items-center">
+                            <i class="fas fa-server text-[var(--color-primary)] mr-2 text-xs"></i>Backends
+                        </a>
+                        <a href="p2p/" class="block text-[var(--color-text-secondary)] hover:text-[var(--color-text-primary)] hover:bg-[var(--color-bg-primary)] px-3 py-2 text-sm transition-colors flex items-center">
+                            <i class="fa-solid fa-circle-nodes text-[var(--color-primary)] mr-2 text-xs"></i>Swarm
+                        </a>
+                        <a href="#" @click.prevent="$store.router.navigate('manage'); settingsOpen = false" class="block text-[var(--color-text-secondary)] hover:text-[var(--color-text-primary)] hover:bg-[var(--color-bg-primary)] px-3 py-2 text-sm transition-colors flex items-center">
+                            <i class="fas fa-cog text-[var(--color-primary)] mr-2 text-xs"></i>System
+                        </a>
+                    </div>
+                </div>
+            </div>
+        </div>
+        
+        <!-- Collapsible menu for small screens -->
+        <div class="lg:hidden" x-show="mobileMenuOpen" x-transition>
+            <div class="pt-3 pb-2 space-y-1 border-t border-[var(--color-bg-secondary)] mt-2">
+                <a href="#" @click.prevent="$store.router.navigate('home'); mobileMenuOpen = false" 
+                   :class="$store.router.currentRoute === 'home' ? 'text-[var(--color-primary)] bg-[var(--color-bg-secondary)]' : 'text-[var(--color-text-secondary)]'"
+                   class="block hover:text-[var(--color-text-primary)] hover:bg-[var(--color-bg-secondary)] px-3 py-2 rounded-lg transition duration-300 ease-in-out flex items-center text-sm">
+                    <i class="fas fa-home text-[var(--color-primary)] mr-3 w-5 text-center text-sm"></i>Home
+                </a>
+                <a href="#" @click.prevent="$store.router.navigate('chat'); mobileMenuOpen = false"
+                   :class="$store.router.currentRoute === 'chat' ? 'text-[var(--color-primary)] bg-[var(--color-bg-secondary)]' : 'text-[var(--color-text-secondary)]'"
+                   class="block hover:text-[var(--color-text-primary)] hover:bg-[var(--color-bg-secondary)] px-3 py-2 rounded-lg transition duration-300 ease-in-out flex items-center text-sm">
+                    <i class="fa-solid fa-comments text-[var(--color-primary)] mr-3 w-5 text-center text-sm"></i>Chat
+                </a>
+                <a href="#" @click.prevent="$store.router.navigate('text2image'); mobileMenuOpen = false"
+                   :class="$store.router.currentRoute === 'text2image' ? 'text-[var(--color-primary)] bg-[var(--color-bg-secondary)]' : 'text-[var(--color-text-secondary)]'"
+                   class="block hover:text-[var(--color-text-primary)] hover:bg-[var(--color-bg-secondary)] px-3 py-2 rounded-lg transition duration-300 ease-in-out flex items-center text-sm">
+                    <i class="fas fa-image text-[var(--color-primary)] mr-3 w-5 text-center text-sm"></i>Images
+                </a>
+                <a href="#" @click.prevent="$store.router.navigate('tts'); mobileMenuOpen = false"
+                   :class="$store.router.currentRoute === 'tts' ? 'text-[var(--color-primary)] bg-[var(--color-bg-secondary)]' : 'text-[var(--color-text-secondary)]'"
+                   class="block hover:text-[var(--color-text-primary)] hover:bg-[var(--color-bg-secondary)] px-3 py-2 rounded-lg transition duration-300 ease-in-out flex items-center text-sm">
+                    <i class="fa-solid fa-music text-[var(--color-primary)] mr-3 w-5 text-center text-sm"></i>TTS
+                </a>
+                <a href="#" @click.prevent="$store.router.navigate('talk'); mobileMenuOpen = false"
+                   :class="$store.router.currentRoute === 'talk' ? 'text-[var(--color-primary)] bg-[var(--color-bg-secondary)]' : 'text-[var(--color-text-secondary)]'"
+                   class="block hover:text-[var(--color-text-primary)] hover:bg-[var(--color-bg-secondary)] px-3 py-2 rounded-lg transition duration-300 ease-in-out flex items-center text-sm">
+                    <i class="fa-solid fa-phone text-[var(--color-primary)] mr-3 w-5 text-center text-sm"></i>Talk
+                </a>
+                <a href="agent-jobs" class="block text-[var(--color-text-secondary)] hover:text-[var(--color-text-primary)] hover:bg-[var(--color-bg-secondary)] px-3 py-2 rounded-lg transition duration-300 ease-in-out flex items-center text-sm">
+                    <i class="fas fa-tasks text-[var(--color-primary)] mr-3 w-5 text-center text-sm"></i>Agent Jobs
+                </a>
+                <a href="traces/" class="block text-[var(--color-text-secondary)] hover:text-[var(--color-text-primary)] hover:bg-[var(--color-bg-secondary)] px-3 py-2 rounded-lg transition duration-300 ease-in-out flex items-center text-sm">
+                    <i class="fas fa-chart-line text-[var(--color-primary)] mr-3 w-5 text-center text-sm"></i>Traces
+                </a>
+                <a href="swagger/index.html" class="block text-[var(--color-text-secondary)] hover:text-[var(--color-text-primary)] hover:bg-[var(--color-bg-secondary)] px-3 py-2 rounded-lg transition duration-300 ease-in-out flex items-center text-sm">
+                    <i class="fas fa-code text-[var(--color-primary)] mr-3 w-5 text-center text-sm"></i>API
+                </a>
+                
+                <!-- System with submenu -->
+                <div>
+                    <button @click="mobileSettingsOpen = !mobileSettingsOpen" 
+                            class="w-full text-left text-[var(--color-text-secondary)] hover:text-[var(--color-text-primary)] hover:bg-[var(--color-bg-secondary)] px-3 py-2 rounded-lg transition duration-300 ease-in-out flex items-center justify-between text-sm">
+                        <div class="flex items-center">
+                            <i class="fas fa-cog text-[var(--color-primary)] mr-3 w-5 text-center text-sm"></i>Settings
+                        </div>
+                        <i class="fas fa-chevron-down text-xs transition-transform" :class="mobileSettingsOpen ? 'rotate-180' : ''"></i>
+                    </button>
+                    <div x-show="mobileSettingsOpen" x-transition class="overflow-hidden">
+                        <a href="#" @click.prevent="$store.router.navigate('browse'); mobileMenuOpen = false" class="block text-[var(--color-text-secondary)] hover:text-[var(--color-text-primary)] hover:bg-[var(--color-bg-secondary)] pl-8 pr-3 py-2 rounded-lg transition duration-300 ease-in-out flex items-center text-sm">
+                            <i class="fas fa-brain text-[var(--color-primary)] mr-3 w-5 text-center text-xs"></i>Models
+                        </a>
+                        <a href="browse/backends" class="block text-[var(--color-text-secondary)] hover:text-[var(--color-text-primary)] hover:bg-[var(--color-bg-secondary)] pl-8 pr-3 py-2 rounded-lg transition duration-300 ease-in-out flex items-center text-sm">
+                            <i class="fas fa-server text-[var(--color-primary)] mr-3 w-5 text-center text-xs"></i>Backends
+                        </a>
+                        <a href="p2p/" class="block text-[var(--color-text-secondary)] hover:text-[var(--color-text-primary)] hover:bg-[var(--color-bg-secondary)] pl-8 pr-3 py-2 rounded-lg transition duration-300 ease-in-out flex items-center text-sm">
+                            <i class="fa-solid fa-circle-nodes text-[var(--color-primary)] mr-3 w-5 text-center text-xs"></i>Swarm
+                        </a>
+                        <a href="#" @click.prevent="$store.router.navigate('manage'); mobileMenuOpen = false" class="block text-[var(--color-text-secondary)] hover:text-[var(--color-text-primary)] hover:bg-[var(--color-bg-secondary)] pl-8 pr-3 py-2 rounded-lg transition duration-300 ease-in-out flex items-center text-sm">
+                            <i class="fas fa-cog text-[var(--color-primary)] mr-3 w-5 text-center text-xs"></i>System
+                        </a>
+                    </div>
+                </div>
+            </div>
+        </div>
+    </div>
+</nav>
--- a/core/http/views/settings.html
+++ b/core/http/views/settings.html
@@ -485,28 +485,6 @@
                </div>
            </div>

-            <!-- Open Responses Settings Section -->
-            <div class="bg-[var(--color-bg-secondary)] border border-[var(--color-accent)]/20 rounded-lg p-6">
-                <h2 class="text-xl font-semibold text-[var(--color-text-primary)] mb-4 flex items-center">
-                    <i class="fas fa-database mr-2 text-[var(--color-accent)] text-sm"></i>
-                    Open Responses Settings
-                </h2>
-                <p class="text-xs text-[var(--color-text-secondary)] mb-4">
-                    Configure Open Responses API response storage
-                </p>
-
-                <div class="space-y-4">
-                    <!-- Store TTL -->
-                    <div>
-                        <label class="block text-sm font-medium text-[var(--color-text-primary)] mb-2">Response Store TTL</label>
-                        <p class="text-xs text-[var(--color-text-secondary)] mb-2">Time-to-live for stored responses (e.g., 1h, 30m, 0 = no expiration)</p>
-                        <input type="text" x-model="settings.open_responses_store_ttl"
-                               placeholder="0"
-                               class="w-full px-3 py-2 bg-[var(--color-bg-primary)] border border-[var(--color-accent)]/20 rounded text-sm text-[var(--color-text-primary)] focus:outline-none focus:ring-2 focus:ring-[var(--color-accent)]/50">
-                    </div>
-                </div>
-            </div>
-
            <!-- API Keys Settings Section -->
            <div class="bg-[var(--color-bg-secondary)] border border-[var(--color-error-light)] rounded-lg p-6">
                <h2 class="text-xl font-semibold text-[var(--color-text-primary)] mb-4 flex items-center">
@@ -655,8 +633,7 @@ function settingsDashboard() {
            galleries_json: '[]',
            backend_galleries_json: '[]',
            api_keys_text: '',
-            agent_job_retention_days: 30,
-            open_responses_store_ttl: '0'
+            agent_job_retention_days: 30
        },
        sourceInfo: '',
        saving: false,
@@ -703,8 +680,7 @@ function settingsDashboard() {
                        galleries_json: JSON.stringify(data.galleries || [], null, 2),
                        backend_galleries_json: JSON.stringify(data.backend_galleries || [], null, 2),
                        api_keys_text: (data.api_keys || []).join('\n'),
-                        agent_job_retention_days: data.agent_job_retention_days || 30,
-                        open_responses_store_ttl: data.open_responses_store_ttl || '0'
+                        agent_job_retention_days: data.agent_job_retention_days || 30
                    };
                    this.sourceInfo = data.source || 'default';
                } else {
@@ -862,9 +838,6 @@ function settingsDashboard() {
                if (this.settings.agent_job_retention_days !== undefined) {
                    payload.agent_job_retention_days = parseInt(this.settings.agent_job_retention_days) || 30;
                }
-                if (this.settings.open_responses_store_ttl !== undefined) {
-                    payload.open_responses_store_ttl = this.settings.open_responses_store_ttl;
-                }

                const response = await fetch('/api/settings', {
                    method: 'POST',
--- a/Show More
+++ b/Show More
Author	SHA1	Message	Date
copilot-swe-agent[bot]	5ecda78be4	Fix: Move Alpine.js router store registration inline The spa-router.js was loaded with defer but registered the Alpine.js store using the 'alpine:init' event. Since Alpine.js also loads with defer, there was a race condition where Alpine could initialize before the event listener was registered, causing $store.router to be undefined. Moved the entire router store definition and registration inline in spa.html so it's guaranteed to be registered before Alpine.js initializes. Co-authored-by: mudler <2420543+mudler@users.noreply.github.com>	2026-01-11 08:34:47 +00:00
copilot-swe-agent[bot]	8da5ef7231	Fix: Remove talk.js from SPA to prevent null element errors The talk.js script has global-scope DOM access (getElementById at top level) that causes "recordButton is null" errors when loaded in the SPA context. The SPA talk view has its own inline JavaScript that handles recording functionality with properly namespaced element IDs (spa-recordButton, etc.). Co-authored-by: mudler <2420543+mudler@users.noreply.github.com>	2026-01-11 08:10:21 +00:00
copilot-swe-agent[bot]	4758996936	Fix Alpine.js component functions not being defined Move critical Alpine.js component functions (resourceMonitor, homeInputForm, startChatSPA, stopModel, stopAllModels, formatBytes) from deferred scripts to inline script block in spa.html. This ensures these functions are defined before Alpine.js processes the DOM and attempts to evaluate x-data expressions. Co-authored-by: mudler <2420543+mudler@users.noreply.github.com>	2026-01-11 07:52:04 +00:00
copilot-swe-agent[bot]	9a50215867	Fix template error: remove invalid .Version reference in manage.html SystemBackend struct does not have a Version field. Updated the backends section to display IsSystem and IsMeta badges instead, matching the original manage.html template. Co-authored-by: mudler <2420543+mudler@users.noreply.github.com>	2026-01-11 07:38:52 +00:00
copilot-swe-agent[bot]	4435c8af57	Fix code review issues in SPA views - Fix text2image to use genImage function from image.js - Add @change handler to sync model select with hidden input - Fix TTS to sync select with hidden input - Simplify TTS model retrieval logic Co-authored-by: mudler <2420543+mudler@users.noreply.github.com>	2026-01-10 23:54:42 +00:00
copilot-swe-agent[bot]	65a57daba6	Convert webui to single-page Alpine.js app - Create SPA container (spa.html) with Alpine.js routing - Create view partials for home, chat, text2image, tts, talk, manage, and browse views - Create spa-router.js for client-side navigation - Create spa-home.js with home view Alpine.js components - Create spa_navbar.html with SPA-aware navigation - Update welcome endpoint to serve SPA instead of separate pages - Update UI routes to serve SPA for chat, text2image, tts, and talk routes Co-authored-by: mudler <2420543+mudler@users.noreply.github.com>	2026-01-10 23:52:00 +00:00
copilot-swe-agent[bot]	b5465cbc3a	Initial plan	2026-01-10 23:28:40 +00:00