chore(model-gallery): ⬆️ update checksum (#10469)

⬆️ Checksum updates in gallery/index.yaml

Signed-off-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
Co-authored-by: mudler <2420543+mudler@users.noreply.github.com>

.github/workflows/backend.yml

@@ -44,7 +44,7 @@ jobs:
       has-merges-singlearch: ${{ steps.set-matrix.outputs['has-merges-singlearch'] }}
     steps:
       - name: Checkout repository
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
 
       - name: Setup Bun
         uses: oven-sh/setup-bun@v2

.github/workflows/backend_build.yml

@@ -101,7 +101,7 @@ jobs:
     steps:
 
       - name: Checkout
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           submodules: true
 

.github/workflows/backend_build_darwin.yml

@@ -57,7 +57,7 @@ jobs:
       HOMEBREW_NO_ANALYTICS: '1'
     steps:
       - name: Clone
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           submodules: true
 

.github/workflows/backend_merge.yml

@@ -49,7 +49,7 @@ jobs:
       # Sparse checkout: the merge job needs `.github/scripts/` (for the
       # keepalive cleanup script) but none of the source tree.
       - name: Checkout (.github/scripts only)
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           sparse-checkout: |
             .github/scripts

.github/workflows/backend_pr.yml

@@ -23,7 +23,7 @@ jobs:
       has-merges-singlearch: ${{ steps.set-matrix.outputs['has-merges-singlearch'] }}
     steps:
       - name: Checkout repository
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
 
       - name: Setup Bun
         uses: oven-sh/setup-bun@v2

.github/workflows/base-images.yml

@@ -127,7 +127,7 @@ jobs:
             # the original l4t matrix entry which set skip-drivers: 'true'.
             skip-drivers: 'true'
     steps:
-      - uses: actions/checkout@v6
+      - uses: actions/checkout@v7
         with:
           submodules: false
       - name: Free disk space

.github/workflows/build-test.yaml

@@ -11,7 +11,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           fetch-depth: 0
       - name: Set up Go
@@ -25,7 +25,7 @@ jobs:
     runs-on: macos-latest
     steps:
       - name: Checkout
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           fetch-depth: 0
       - name: Set up Go
@@ -47,7 +47,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           fetch-depth: 0
       - name: Configure apt mirror on runner

.github/workflows/bump-inference-defaults.yml

@@ -14,7 +14,7 @@ jobs:
   bump:
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v6
+      - uses: actions/checkout@v7
 
       - uses: actions/setup-go@v5
         with:

.github/workflows/bump_deps.yaml

@@ -92,7 +92,7 @@ jobs:
             file: "backend/go/vibevoice-cpp/Makefile"
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v6
+      - uses: actions/checkout@v7
       - name: Bump dependencies 🔧
         id: bump
         run: |
@@ -128,7 +128,7 @@ jobs:
     if: github.repository == 'mudler/LocalAI'
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v6
+      - uses: actions/checkout@v7
       - name: Bump vLLM cu130 wheel pin 🔧
         id: bump
         run: |

.github/workflows/bump_docs.yaml

@@ -13,7 +13,7 @@ jobs:
           - repository: "mudler/LocalAI"
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v6
+      - uses: actions/checkout@v7
       - name: Bump dependencies 🔧
         run: |
           bash .github/bump_docs.sh ${{ matrix.repository }}

.github/workflows/checksum_checker.yaml

@@ -8,7 +8,7 @@ jobs:
     if: github.repository == 'mudler/LocalAI'
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v6
+      - uses: actions/checkout@v7
       - name: Configure apt mirror on runner
         uses: ./.github/actions/configure-apt-mirror
       - name: Install dependencies

.github/workflows/deploy-explorer.yaml

@@ -16,7 +16,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Clone
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           submodules: true
       - uses: actions/setup-go@v5

.github/workflows/gallery-agent.yaml

@@ -31,7 +31,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout repository
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           token: ${{ secrets.GITHUB_TOKEN }}
 

.github/workflows/generate_intel_image.yaml

@@ -44,7 +44,7 @@ jobs:
         uses: docker/setup-buildx-action@master
 
       - name: Checkout
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
 
       - name: Cache Intel images
         uses: docker/build-push-action@v7

.github/workflows/gh-pages.yml

@@ -28,7 +28,7 @@ jobs:
       HUGO_VERSION: "0.146.3"
     steps:
       - name: Checkout
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           fetch-depth: 0  # needed for enableGitInfo
           submodules: true

.github/workflows/image_build.yml

@@ -80,7 +80,7 @@ jobs:
     steps:
 
       - name: Checkout
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
 
       - name: Configure apt mirror on runner
         id: apt_mirror

.github/workflows/image_merge.yml

@@ -36,7 +36,7 @@ jobs:
       # Sparse checkout: needed for .github/scripts/ (the keepalive cleanup
       # script). Skips the rest of the source tree.
       - name: Checkout (.github/scripts only)
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           sparse-checkout: |
             .github/scripts

.github/workflows/lint.yml

@@ -20,7 +20,7 @@ jobs:
   golangci-lint:
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v6
+      - uses: actions/checkout@v7
         with:
           # Full history so golangci-lint's new-from-merge-base can reach
           # origin/master and compute the diff against it.

.github/workflows/release.yaml

@@ -10,7 +10,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           fetch-depth: 0
       - name: Set up Go
@@ -28,7 +28,7 @@ jobs:
     runs-on: macos-latest
     steps:
       - name: Checkout
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           fetch-depth: 0
       - name: Set up Go
@@ -46,7 +46,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           fetch-depth: 0
       - name: Configure apt mirror on runner

.github/workflows/secscan.yaml

@@ -14,7 +14,7 @@ jobs:
       GO111MODULE: on
     steps:
       - name: Checkout Source
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         if: ${{ github.actor != 'dependabot[bot]' }}
       - name: Run Gosec Security Scanner
         if: ${{ github.actor != 'dependabot[bot]' }}

.github/workflows/test-extra.yml

@@ -50,7 +50,7 @@ jobs:
       parakeet-cpp: ${{ steps.detect.outputs.parakeet-cpp }}
     steps:
       - name: Checkout repository
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
       - name: Setup Bun
         uses: oven-sh/setup-bun@v2
       - name: Install dependencies
@@ -67,7 +67,7 @@ jobs:
   #   runs-on: ubuntu-latest
   #   steps:
   #     - name: Clone
-  #       uses: actions/checkout@v6
+  #       uses: actions/checkout@v7
   #       with:
   #         submodules: true
   #     - name: Dependencies
@@ -90,7 +90,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Clone
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           submodules: true
       - name: Dependencies
@@ -113,7 +113,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Clone
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           submodules: true
       - name: Dependencies
@@ -137,7 +137,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Clone
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           submodules: true
       - name: Dependencies
@@ -158,7 +158,7 @@ jobs:
   #  runs-on: ubuntu-latest
   #  steps:
   #    - name: Clone
-  #      uses: actions/checkout@v6
+  #      uses: actions/checkout@v7
   #      with:
   #        submodules: true
   #    - name: Dependencies
@@ -178,7 +178,7 @@ jobs:
   #   runs-on: ubuntu-latest
   #   steps:
   #     - name: Clone
-  #       uses: actions/checkout@v6
+  #       uses: actions/checkout@v7
   #       with:
   #         submodules: true
   #     - name: Dependencies
@@ -240,7 +240,7 @@ jobs:
   #           sudo rm -rf "$AGENT_TOOLSDIRECTORY" || true
   #           df -h
   #     - name: Clone
-  #       uses: actions/checkout@v6
+  #       uses: actions/checkout@v7
   #       with:
   #         submodules: true
   #     - name: Dependencies
@@ -265,7 +265,7 @@ jobs:
   #   runs-on: ubuntu-latest
   #   steps:
   #     - name: Clone
-  #       uses: actions/checkout@v6
+  #       uses: actions/checkout@v7
   #       with:
   #         submodules: true
   #     - name: Dependencies
@@ -288,7 +288,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Clone
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           submodules: true
       - name: Dependencies
@@ -309,7 +309,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Clone
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           submodules: true
       - name: Dependencies
@@ -330,7 +330,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Clone
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           submodules: true
       - name: Dependencies
@@ -351,7 +351,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Clone
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           submodules: true
       - name: Dependencies
@@ -373,7 +373,7 @@ jobs:
   #   timeout-minutes: 45
   #   steps:
   #     - name: Clone
-  #       uses: actions/checkout@v6
+  #       uses: actions/checkout@v7
   #       with:
   #         submodules: true
   #     - name: Dependencies
@@ -394,7 +394,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Clone
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           submodules: true
       - name: Dependencies
@@ -415,7 +415,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Clone
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           submodules: true
       - name: Dependencies
@@ -436,7 +436,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Clone
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           submodules: true
       - name: Dependencies
@@ -462,7 +462,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Clone
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           submodules: true
       - name: Dependencies
@@ -484,7 +484,7 @@ jobs:
     timeout-minutes: 30
     steps:
       - name: Clone
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           submodules: true
       - name: Dependencies
@@ -513,7 +513,7 @@ jobs:
     timeout-minutes: 90
     steps:
       - name: Clone
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           submodules: true
       - name: Setup Go
@@ -530,7 +530,7 @@ jobs:
     timeout-minutes: 90
     steps:
       - name: Clone
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           submodules: true
       - name: Setup Go
@@ -552,7 +552,7 @@ jobs:
     timeout-minutes: 20
     steps:
       - name: Clone
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           submodules: true
       - name: Setup Go
@@ -579,7 +579,7 @@ jobs:
     timeout-minutes: 90
     steps:
       - name: Clone
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           submodules: true
       - name: Setup Go
@@ -604,7 +604,7 @@ jobs:
     timeout-minutes: 90
     steps:
       - name: Clone
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           submodules: true
       - name: Setup Go
@@ -625,7 +625,7 @@ jobs:
     timeout-minutes: 90
     steps:
       - name: Clone
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           submodules: true
       - name: Setup Go
@@ -645,7 +645,7 @@ jobs:
     timeout-minutes: 90
     steps:
       - name: Clone
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           submodules: true
       - name: Setup Go
@@ -664,7 +664,7 @@ jobs:
     timeout-minutes: 90
     steps:
       - name: Clone
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           submodules: true
       - name: Setup Go
@@ -681,7 +681,7 @@ jobs:
     timeout-minutes: 90
     steps:
       - name: Clone
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           submodules: true
       - name: Setup Go
@@ -698,7 +698,7 @@ jobs:
     timeout-minutes: 90
     steps:
       - name: Clone
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           submodules: true
       - name: Setup Go
@@ -741,7 +741,7 @@ jobs:
   #   timeout-minutes: 90
   #   steps:
   #     - name: Clone
-  #       uses: actions/checkout@v6
+  #       uses: actions/checkout@v7
   #       with:
   #         submodules: true
   #     - name: Dependencies
@@ -783,7 +783,7 @@ jobs:
   #   timeout-minutes: 90
   #   steps:
   #     - name: Clone
-  #       uses: actions/checkout@v6
+  #       uses: actions/checkout@v7
   #       with:
   #         submodules: true
   #     - name: Dependencies
@@ -808,7 +808,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Clone
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           submodules: true
       - name: Dependencies
@@ -840,7 +840,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Clone
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           submodules: true
       - name: Dependencies
@@ -876,7 +876,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Clone
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           submodules: true
       - name: Dependencies
@@ -915,7 +915,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Clone
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           submodules: true
       - name: Dependencies
@@ -952,7 +952,7 @@ jobs:
     timeout-minutes: 90
     steps:
       - name: Clone
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           submodules: true
       - name: Dependencies
@@ -987,7 +987,7 @@ jobs:
     timeout-minutes: 90
     steps:
       - name: Clone
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           submodules: true
       - name: Setup Go
@@ -1013,7 +1013,7 @@ jobs:
     timeout-minutes: 150
     steps:
       - name: Clone
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           submodules: true
       - name: Dependencies
@@ -1042,7 +1042,7 @@ jobs:
     timeout-minutes: 60
     steps:
       - name: Clone
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           submodules: true
       - name: Setup Go
@@ -1058,7 +1058,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Clone
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           submodules: true
       - name: Dependencies
@@ -1091,7 +1091,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Clone
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           submodules: true
       - name: Dependencies
@@ -1114,7 +1114,7 @@ jobs:
     timeout-minutes: 90
     steps:
       - name: Clone
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           submodules: true
       - name: Dependencies
@@ -1140,7 +1140,7 @@ jobs:
     timeout-minutes: 90
     steps:
       - name: Clone
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           submodules: true
       - name: Dependencies

.github/workflows/test.yml

@@ -21,7 +21,7 @@ jobs:
         go-version: ['1.26.x']
     steps:
       - name: Clone
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           submodules: true
       - name: Free disk space
@@ -84,7 +84,7 @@ jobs:
         go-version: ['1.26.x']
     steps:
       - name: Clone
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           submodules: true
       - name: Setup Go ${{ matrix.go-version }}

.github/workflows/tests-aio.yml

@@ -62,7 +62,7 @@ jobs:
           sudo rm -rfv build || true
           df -h
       - name: Clone
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           submodules: true
       - name: Dependencies

.github/workflows/tests-e2e.yml

@@ -21,7 +21,7 @@ jobs:
         go-version: ['1.25.x']
     steps:
       - name: Clone
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           submodules: true
       - name: Configure apt mirror on runner

.github/workflows/tests-pii-ner-e2e.yml

@@ -0,0 +1,97 @@
+---
+name: 'PII NER tier E2E (live GGUF, CPU)'
+
+# Runs the real privacy-filter GGUF NER tier end-to-end on CPU — the gap the
+# hermetic tests/e2e suite cannot cover (it only exercises the in-process
+# pattern tier). Heavy (builds the C++ backend image + downloads a ~2.7 GB
+# GGUF), so it is path-filtered on PRs and otherwise runs nightly / on demand.
+#
+# This drives the container-level harness (tests/e2e-backends) via
+# `make test-extra-backend-privacy-filter`: it builds the privacy-filter image,
+# downloads the model, loads it on CPU, and asserts byte-correct, UTF-8-aligned
+# TokenClassify spans. The complementary HTTP-path specs in tests/e2e
+# (e2e_pii_ner_test.go) Skip unless PII_NER_MODEL_GGUF is wired.
+
+on:
+  workflow_dispatch:
+  schedule:
+    - cron: '0 3 * * *'
+  push:
+    branches:
+      - master
+    paths:
+      - 'backend/cpp/privacy-filter/**'
+      - 'backend/Dockerfile.privacy-filter'
+      - 'core/services/routing/pii/**'
+      - 'core/services/routing/piidetector/**'
+      - 'core/backend/token_classify.go'
+      - 'core/http/endpoints/localai/pii.go'
+      - 'core/schema/pii.go'
+      - 'tests/e2e-backends/**'
+      - 'tests/e2e/e2e_pii_ner_test.go'
+      - 'tests/e2e/e2e_suite_test.go'
+      - '.github/workflows/tests-pii-ner-e2e.yml'
+  pull_request:
+    paths:
+      - 'backend/cpp/privacy-filter/**'
+      - 'backend/Dockerfile.privacy-filter'
+      - 'core/services/routing/pii/**'
+      - 'core/services/routing/piidetector/**'
+      - 'core/backend/token_classify.go'
+      - 'core/http/endpoints/localai/pii.go'
+      - 'core/schema/pii.go'
+      - 'tests/e2e-backends/**'
+      - 'tests/e2e/e2e_pii_ner_test.go'
+      - 'tests/e2e/e2e_suite_test.go'
+      - '.github/workflows/tests-pii-ner-e2e.yml'
+
+concurrency:
+  group: ci-tests-pii-ner-e2e-${{ github.event.pull_request.number || github.sha }}-${{ github.repository }}
+  cancel-in-progress: ${{ github.event_name == 'pull_request' }}
+
+jobs:
+  tests-pii-ner-e2e:
+    runs-on: ubuntu-latest
+    strategy:
+      matrix:
+        go-version: ['1.25.x']
+    steps:
+      - name: Clone
+        uses: actions/checkout@v7
+        with:
+          submodules: true
+      - name: Free disk space
+        run: |
+          sudo rm -rf /usr/share/dotnet /usr/local/lib/android /opt/ghc /opt/hostedtoolcache/CodeQL || true
+          sudo docker image prune --all --force || true
+          df -h
+      - name: Configure apt mirror on runner
+        uses: ./.github/actions/configure-apt-mirror
+      - name: Setup Go ${{ matrix.go-version }}
+        uses: actions/setup-go@v5
+        with:
+          go-version: ${{ matrix.go-version }}
+          cache: false
+      - name: Proto Dependencies
+        run: |
+          curl -L -s https://github.com/protocolbuffers/protobuf/releases/download/v26.1/protoc-26.1-linux-x86_64.zip -o protoc.zip && \
+          unzip -j -d /usr/local/bin protoc.zip bin/protoc && \
+          rm protoc.zip
+          go install google.golang.org/protobuf/cmd/protoc-gen-go@v1.34.2
+          go install google.golang.org/grpc/cmd/protoc-gen-go-grpc@1958fcbe2ca8bd93af633f11e97d44e567e945af
+          PATH="$PATH:$HOME/go/bin" make protogen-go
+      - name: Dependencies
+        run: |
+          sudo apt-get update
+          sudo apt-get install -y build-essential
+      # Builds local-ai-backend:privacy-filter, downloads the GGUF, loads it on
+      # CPU and runs the token_classify capability spec (byte-offset contract).
+      - name: Run live PII NER backend E2E
+        run: PATH="$PATH:$HOME/go/bin" make test-extra-backend-privacy-filter
+      - name: Setup tmate session if tests fail
+        if: ${{ failure() }}
+        uses: mxschmitt/action-tmate@v3.23
+        with:
+          detached: true
+          connect-timeout-seconds: 180
+          limit-access-to-actor: true

.github/workflows/tests-ui-e2e.yml

@@ -23,7 +23,7 @@ jobs:
         go-version: ['1.26.x']
     steps:
       - name: Clone
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           submodules: true
       - name: Configure apt mirror on runner

.github/workflows/update_swagger.yaml

@@ -10,7 +10,7 @@ jobs:
       fail-fast: false
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v6
+      - uses: actions/checkout@v7
       - name: Configure apt mirror on runner
         uses: ./.github/actions/configure-apt-mirror
       - uses: actions/setup-go@v5

.gitignore

@@ -91,3 +91,6 @@ core/http/react-ui/test-results/
 
 # Local worktrees
 .worktrees/
+
+# SDD / brainstorm scratch (agent-driven development)
+.superpowers/

Makefile

@@ -690,6 +690,16 @@ test-extra-backend-llama-cpp-transcription: docker-build-llama-cpp
 	BACKEND_TEST_CTX_SIZE=2048 \
 	$(MAKE) test-extra-backend
 
+## privacy-filter: the PII/NER token-classification backend. Exercises the
+## TokenClassify RPC and asserts byte-correct, UTF-8-aligned span offsets
+## against the openai-privacy-filter multilingual GGUF (CPU-runnable, ~50M
+## active params). This is the live-backend coverage for the PII NER tier.
+test-extra-backend-privacy-filter: docker-build-privacy-filter
+	BACKEND_IMAGE=local-ai-backend:privacy-filter \
+	BACKEND_TEST_MODEL_URL=https://huggingface.co/LocalAI-io/privacy-filter-multilingual-GGUF/resolve/main/privacy-filter-multilingual-f16.gguf \
+	BACKEND_TEST_CAPS=health,load,token_classify \
+	$(MAKE) test-extra-backend
+
 ## vllm is resolved from a HuggingFace model id (no file download) and
 ## exercises Predict + streaming + tool-call extraction via the hermes parser.
 ## Requires a host CPU with the SIMD instructions the prebuilt vllm CPU

backend/cpp/llama-cpp/Makefile

@@ -1,5 +1,5 @@
 
-LLAMA_VERSION?=e475fa2b5f9fb50c3d6fc3e7c6fdf1e004465b62
+LLAMA_VERSION?=73618f27a801c0b8614ceaf3547d3c2a99baae14
 LLAMA_REPO?=https://github.com/ggerganov/llama.cpp
 
 CMAKE_ARGS?=

backend/go/crispasr/Makefile

@@ -8,7 +8,7 @@ JOBS?=$(shell nproc --ignore=1)
 
 # CrispASR version (release tag)
 CRISPASR_REPO?=https://github.com/CrispStrobe/CrispASR
-CRISPASR_VERSION?=d745bda4386ae0f9d1d2f23fff8ec95d76428221
+CRISPASR_VERSION?=63b57289255267edf66e43e33bc3911e04a2e92d
 SO_TARGET?=libgocrispasr.so
 
 CMAKE_ARGS+=-DBUILD_SHARED_LIBS=OFF

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?=b12098f5d09fc83da36e65c784f7bdb16a5a5ebf
+STABLEDIFFUSION_GGML_VERSION?=f440ad9c29dd8bc34e5d1f4b863832b96d6ea05f
 
 CMAKE_ARGS+=-DGGML_MAX_NAME=128
 

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?=5ed76e9a079962f1c85cfce44edd325c27ef1f97
+WHISPER_CPP_VERSION?=bae6bc02b1940bbfb87b6a0299c565e563b916d1
 SO_TARGET?=libgowhisper.so
 
 CMAKE_ARGS+=-DBUILD_SHARED_LIBS=OFF

backend/python/diffusers/requirements-cpu.txt

@@ -1,7 +1,7 @@
 --extra-index-url https://download.pytorch.org/whl/cpu
-git+https://github.com/huggingface/diffusers
+diffusers==0.38.0
 opencv-python
-transformers
+transformers==4.57.6
 torchvision==0.22.1
 accelerate
 git+https://github.com/xhinker/sd_embed
@@ -10,9 +10,15 @@ sentencepiece
 torch==2.7.1
 optimum-quanto
 ftfy
-# TODO: re-add compel once it supports transformers >= 5.
-# Tracking: https://github.com/damian0815/compel/pull/129
-#           https://github.com/damian0815/compel/issues/128
-# compel currently pins transformers~=4.25, which forced pip into multi-hour
-# resolver backtracking storms in CI. backend.py imports it lazily and gates
-# the COMPEL=1 env var on the import succeeding, so dropping it here is safe.
+# diffusers and transformers are pinned together on purpose. transformers v5
+# restructured CLIPTextModel and dropped the `.text_model` attribute, which
+# breaks single-file Stable Diffusion loading on every released diffusers
+# (<=0.38.0); only unreleased diffusers main supports transformers v5. Tracking
+# main via git froze whichever broken pair existed at image-build time. Pin the
+# last known-good released pair so builds are reproducible and can't drift into
+# the broken window. See https://github.com/mudler/LocalAI/issues/9979
+#
+# compel is intentionally omitted: it pins transformers~=4.25, which conflicts
+# with this pin and previously forced pip into multi-hour resolver backtracking
+# storms in CI. backend.py imports it lazily and gates the COMPEL=1 env var on
+# the import succeeding, so dropping it here is safe.

backend/python/diffusers/requirements-cublas12.txt

@@ -1,7 +1,7 @@
 --extra-index-url https://download.pytorch.org/whl/cu121
-git+https://github.com/huggingface/diffusers
+diffusers==0.38.0
 opencv-python
-transformers
+transformers==4.57.6
 torchvision
 accelerate
 git+https://github.com/xhinker/sd_embed
@@ -10,9 +10,15 @@ sentencepiece
 torch
 ftfy
 optimum-quanto
-# TODO: re-add compel once it supports transformers >= 5.
-# Tracking: https://github.com/damian0815/compel/pull/129
-#           https://github.com/damian0815/compel/issues/128
-# compel currently pins transformers~=4.25, which forced pip into multi-hour
-# resolver backtracking storms in CI. backend.py imports it lazily and gates
-# the COMPEL=1 env var on the import succeeding, so dropping it here is safe.
+# diffusers and transformers are pinned together on purpose. transformers v5
+# restructured CLIPTextModel and dropped the `.text_model` attribute, which
+# breaks single-file Stable Diffusion loading on every released diffusers
+# (<=0.38.0); only unreleased diffusers main supports transformers v5. Tracking
+# main via git froze whichever broken pair existed at image-build time. Pin the
+# last known-good released pair so builds are reproducible and can't drift into
+# the broken window. See https://github.com/mudler/LocalAI/issues/9979
+#
+# compel is intentionally omitted: it pins transformers~=4.25, which conflicts
+# with this pin and previously forced pip into multi-hour resolver backtracking
+# storms in CI. backend.py imports it lazily and gates the COMPEL=1 env var on
+# the import succeeding, so dropping it here is safe.

backend/python/diffusers/requirements-cublas13.txt

@@ -1,7 +1,7 @@
 --extra-index-url https://download.pytorch.org/whl/cu130
-git+https://github.com/huggingface/diffusers
+diffusers==0.38.0
 opencv-python
-transformers
+transformers==4.57.6
 torchvision
 accelerate
 git+https://github.com/xhinker/sd_embed
@@ -10,9 +10,15 @@ sentencepiece
 torch
 ftfy
 optimum-quanto
-# TODO: re-add compel once it supports transformers >= 5.
-# Tracking: https://github.com/damian0815/compel/pull/129
-#           https://github.com/damian0815/compel/issues/128
-# compel currently pins transformers~=4.25, which forced pip into multi-hour
-# resolver backtracking storms in CI. backend.py imports it lazily and gates
-# the COMPEL=1 env var on the import succeeding, so dropping it here is safe.
+# diffusers and transformers are pinned together on purpose. transformers v5
+# restructured CLIPTextModel and dropped the `.text_model` attribute, which
+# breaks single-file Stable Diffusion loading on every released diffusers
+# (<=0.38.0); only unreleased diffusers main supports transformers v5. Tracking
+# main via git froze whichever broken pair existed at image-build time. Pin the
+# last known-good released pair so builds are reproducible and can't drift into
+# the broken window. See https://github.com/mudler/LocalAI/issues/9979
+#
+# compel is intentionally omitted: it pins transformers~=4.25, which conflicts
+# with this pin and previously forced pip into multi-hour resolver backtracking
+# storms in CI. backend.py imports it lazily and gates the COMPEL=1 env var on
+# the import succeeding, so dropping it here is safe.

backend/python/diffusers/requirements-hipblas.txt

@@ -1,17 +1,23 @@
 --extra-index-url https://download.pytorch.org/whl/rocm7.0
 torch==2.10.0+rocm7.0
 torchvision==0.25.0+rocm7.0
-git+https://github.com/huggingface/diffusers
+diffusers==0.38.0
 opencv-python
-transformers
+transformers==4.57.6
 accelerate
 peft
 sentencepiece
 optimum-quanto
 ftfy
-# TODO: re-add compel once it supports transformers >= 5.
-# Tracking: https://github.com/damian0815/compel/pull/129
-#           https://github.com/damian0815/compel/issues/128
-# compel currently pins transformers~=4.25, which forced pip into multi-hour
-# resolver backtracking storms in CI. backend.py imports it lazily and gates
-# the COMPEL=1 env var on the import succeeding, so dropping it here is safe.
+# diffusers and transformers are pinned together on purpose. transformers v5
+# restructured CLIPTextModel and dropped the `.text_model` attribute, which
+# breaks single-file Stable Diffusion loading on every released diffusers
+# (<=0.38.0); only unreleased diffusers main supports transformers v5. Tracking
+# main via git froze whichever broken pair existed at image-build time. Pin the
+# last known-good released pair so builds are reproducible and can't drift into
+# the broken window. See https://github.com/mudler/LocalAI/issues/9979
+#
+# compel is intentionally omitted: it pins transformers~=4.25, which conflicts
+# with this pin and previously forced pip into multi-hour resolver backtracking
+# storms in CI. backend.py imports it lazily and gates the COMPEL=1 env var on
+# the import succeeding, so dropping it here is safe.

backend/python/diffusers/requirements-intel.txt

@@ -3,18 +3,24 @@ torch
 torchvision
 optimum[openvino]
 setuptools
-git+https://github.com/huggingface/diffusers
+diffusers==0.38.0
 opencv-python
-transformers
+transformers==4.57.6
 accelerate
 git+https://github.com/xhinker/sd_embed
 peft
 sentencepiece
 optimum-quanto
 ftfy
-# TODO: re-add compel once it supports transformers >= 5.
-# Tracking: https://github.com/damian0815/compel/pull/129
-#           https://github.com/damian0815/compel/issues/128
-# compel currently pins transformers~=4.25, which forced pip into multi-hour
-# resolver backtracking storms in CI. backend.py imports it lazily and gates
-# the COMPEL=1 env var on the import succeeding, so dropping it here is safe.
+# diffusers and transformers are pinned together on purpose. transformers v5
+# restructured CLIPTextModel and dropped the `.text_model` attribute, which
+# breaks single-file Stable Diffusion loading on every released diffusers
+# (<=0.38.0); only unreleased diffusers main supports transformers v5. Tracking
+# main via git froze whichever broken pair existed at image-build time. Pin the
+# last known-good released pair so builds are reproducible and can't drift into
+# the broken window. See https://github.com/mudler/LocalAI/issues/9979
+#
+# compel is intentionally omitted: it pins transformers~=4.25, which conflicts
+# with this pin and previously forced pip into multi-hour resolver backtracking
+# storms in CI. backend.py imports it lazily and gates the COMPEL=1 env var on
+# the import succeeding, so dropping it here is safe.

backend/python/diffusers/requirements-l4t12.txt

@@ -1,7 +1,7 @@
 --extra-index-url https://pypi.jetson-ai-lab.io/jp6/cu129/
 torch
-git+https://github.com/huggingface/diffusers
-transformers
+diffusers==0.38.0
+transformers==4.57.6
 accelerate
 peft
 optimum-quanto
@@ -9,9 +9,15 @@ numpy<2
 sentencepiece
 torchvision
 ftfy
-# TODO: re-add compel once it supports transformers >= 5.
-# Tracking: https://github.com/damian0815/compel/pull/129
-#           https://github.com/damian0815/compel/issues/128
-# compel currently pins transformers~=4.25, which forced pip into multi-hour
-# resolver backtracking storms in CI. backend.py imports it lazily and gates
-# the COMPEL=1 env var on the import succeeding, so dropping it here is safe.
+# diffusers and transformers are pinned together on purpose. transformers v5
+# restructured CLIPTextModel and dropped the `.text_model` attribute, which
+# breaks single-file Stable Diffusion loading on every released diffusers
+# (<=0.38.0); only unreleased diffusers main supports transformers v5. Tracking
+# main via git froze whichever broken pair existed at image-build time. Pin the
+# last known-good released pair so builds are reproducible and can't drift into
+# the broken window. See https://github.com/mudler/LocalAI/issues/9979
+#
+# compel is intentionally omitted: it pins transformers~=4.25, which conflicts
+# with this pin and previously forced pip into multi-hour resolver backtracking
+# storms in CI. backend.py imports it lazily and gates the COMPEL=1 env var on
+# the import succeeding, so dropping it here is safe.

backend/python/diffusers/requirements-l4t13.txt

@@ -1,7 +1,7 @@
 --extra-index-url https://download.pytorch.org/whl/cu130
 torch
-git+https://github.com/huggingface/diffusers
-transformers
+diffusers==0.38.0
+transformers==4.57.6
 accelerate
 peft
 optimum-quanto
@@ -10,9 +10,15 @@ sentencepiece
 torchvision
 ftfy
 chardet
-# TODO: re-add compel once it supports transformers >= 5.
-# Tracking: https://github.com/damian0815/compel/pull/129
-#           https://github.com/damian0815/compel/issues/128
-# compel currently pins transformers~=4.25, which forced pip into multi-hour
-# resolver backtracking storms in CI. backend.py imports it lazily and gates
-# the COMPEL=1 env var on the import succeeding, so dropping it here is safe.
+# diffusers and transformers are pinned together on purpose. transformers v5
+# restructured CLIPTextModel and dropped the `.text_model` attribute, which
+# breaks single-file Stable Diffusion loading on every released diffusers
+# (<=0.38.0); only unreleased diffusers main supports transformers v5. Tracking
+# main via git froze whichever broken pair existed at image-build time. Pin the
+# last known-good released pair so builds are reproducible and can't drift into
+# the broken window. See https://github.com/mudler/LocalAI/issues/9979
+#
+# compel is intentionally omitted: it pins transformers~=4.25, which conflicts
+# with this pin and previously forced pip into multi-hour resolver backtracking
+# storms in CI. backend.py imports it lazily and gates the COMPEL=1 env var on
+# the import succeeding, so dropping it here is safe.

backend/python/diffusers/requirements-mps.txt

@@ -1,16 +1,22 @@
 torch==2.7.1
 torchvision==0.22.1
-git+https://github.com/huggingface/diffusers
+diffusers==0.38.0
 opencv-python
-transformers
+transformers==4.57.6
 accelerate
 peft
 sentencepiece
 optimum-quanto
 ftfy
-# TODO: re-add compel once it supports transformers >= 5.
-# Tracking: https://github.com/damian0815/compel/pull/129
-#           https://github.com/damian0815/compel/issues/128
-# compel currently pins transformers~=4.25, which forced pip into multi-hour
-# resolver backtracking storms in CI. backend.py imports it lazily and gates
-# the COMPEL=1 env var on the import succeeding, so dropping it here is safe.
+# diffusers and transformers are pinned together on purpose. transformers v5
+# restructured CLIPTextModel and dropped the `.text_model` attribute, which
+# breaks single-file Stable Diffusion loading on every released diffusers
+# (<=0.38.0); only unreleased diffusers main supports transformers v5. Tracking
+# main via git froze whichever broken pair existed at image-build time. Pin the
+# last known-good released pair so builds are reproducible and can't drift into
+# the broken window. See https://github.com/mudler/LocalAI/issues/9979
+#
+# compel is intentionally omitted: it pins transformers~=4.25, which conflicts
+# with this pin and previously forced pip into multi-hour resolver backtracking
+# storms in CI. backend.py imports it lazily and gates the COMPEL=1 env var on
+# the import succeeding, so dropping it here is safe.

core/application/application.go

@@ -341,11 +341,9 @@ func (a *Application) ResolvePIIPolicy(cfg *config.ModelConfig) (enabled bool, d
 	}
 	appCfg := a.ApplicationConfig()
 
-	if cfg.PII.Enabled != nil {
-		enabled = *cfg.PII.Enabled
-	} else {
-		enabled = cfg.PIIIsEnabled() // backend default (cloud-proxy)
-	}
+	// PIIIsEnabled already encodes "explicit pii.enabled wins, else backend
+	// default (cloud-proxy)" — the single source of that rule.
+	enabled = cfg.PIIIsEnabled()
 	if !enabled {
 		return false, nil
 	}
@@ -354,7 +352,7 @@ func (a *Application) ResolvePIIPolicy(cfg *config.ModelConfig) (enabled bool, d
 	if len(detectors) == 0 {
 		detectors = append([]string(nil), appCfg.PIIDefaultDetectors...)
 	}
-	return enabled, detectors
+	return true, detectors // enabled is necessarily true past the !enabled guard
 }
 
 // PIIPolicyResolver adapts ResolvePIIPolicy to pii.PolicyResolver for

core/application/distributed.go

@@ -357,6 +357,15 @@ func initDistributed(cfg *config.ApplicationConfig, authDB *gorm.DB, configLoade
 		Pressure:         pressure,
 	})
 
+	// Wire staging-progress broadcasting so file-staging shows up on every
+	// replica, not just the one performing the transfer. Without this, a
+	// /api/operations poll that round-robins onto a peer sees no staging row and
+	// the progress flickers. The origin publishes; peers mirror via the wildcard.
+	router.StagingTracker().SetPublisher(natsClient)
+	if _, err := router.StagingTracker().SubscribeBroadcasts(natsClient); err != nil {
+		xlog.Warn("Failed to subscribe to staging progress broadcasts", "error", err)
+	}
+
 	// Create ReplicaReconciler for auto-scaling model replicas. Adapter +
 	// RegistrationToken feed the state-reconciliation passes: pending op
 	// drain uses the adapter, and model health probes use the token to auth

core/config/meta/registry.go

@@ -537,6 +537,36 @@ func DefaultRegistry() map[string]FieldMetaOverride {
 			Component:   "number",
 			Order:       79,
 		},
+		"pipeline.compaction.enabled": {
+			Section:     "pipeline",
+			Label:       "Compaction Enabled",
+			Description: "Fold conversation items that age out of the live window (Max History Items) into a rolling summary instead of dropping them, so long realtime sessions stay cheap without losing earlier context. Off by default.",
+			Component:   "toggle",
+			Order:       80,
+		},
+		"pipeline.compaction.trigger_items": {
+			Section:     "pipeline",
+			Label:       "Compaction Trigger Items",
+			Description: "High-water mark: once the live conversation exceeds this many items, the overflow above Max History Items is summarized and evicted. Must be greater than Max History Items; defaults to twice it. The gap controls how often summarization runs.",
+			Component:   "number",
+			Order:       81,
+		},
+		"pipeline.compaction.summary_model": {
+			Section:     "pipeline",
+			Label:       "Compaction Summary Model",
+			Description: "Optional smaller/cheaper model used to produce the rolling summary. Empty reuses the pipeline's own LLM. On CPU, a tiny model here keeps compaction from competing with the conversation LLM.",
+			Component:   "input",
+			Advanced:    true,
+			Order:       82,
+		},
+		"pipeline.compaction.max_summary_tokens": {
+			Section:     "pipeline",
+			Label:       "Compaction Max Summary Tokens",
+			Description: "Advisory cap on the rolling summary length (fed to the summarizer prompt). Defaults to 512.",
+			Component:   "number",
+			Advanced:    true,
+			Order:       83,
+		},
 
 		// --- Functions ---
 		"function.grammar.parallel_calls": {

core/config/model_config.go

@@ -641,11 +641,32 @@ type Pipeline struct {
 	// context fills.
 	MaxHistoryItems *int `yaml:"max_history_items,omitempty" json:"max_history_items,omitempty"`
 
+	// Compaction folds conversation items that age out of the live window
+	// (max_history_items) into a rolling summary instead of dropping them, so
+	// long realtime sessions stay cheap without losing earlier context. Nil
+	// (block absent) means disabled, preserving existing behavior.
+	Compaction *PipelineCompaction `yaml:"compaction,omitempty" json:"compaction,omitempty"`
+
 	// VoiceRecognition gates the pipeline behind speaker verification. Nil
 	// (block absent) means no gate, preserving existing behavior.
 	VoiceRecognition *PipelineVoiceRecognition `yaml:"voice_recognition,omitempty" json:"voice_recognition,omitempty"`
 }
 
+// PipelineCompaction configures summarize-then-drop for a realtime pipeline.
+type PipelineCompaction struct {
+	// Enabled turns summarize-then-drop on. Default false.
+	Enabled bool `yaml:"enabled,omitempty" json:"enabled,omitempty"`
+	// TriggerItems is the high-water mark: once live items exceed it, overflow
+	// above max_history_items is summarized and evicted. Must exceed
+	// max_history_items; clamped up if not. Default: 2x max_history_items.
+	TriggerItems int `yaml:"trigger_items,omitempty" json:"trigger_items,omitempty"`
+	// SummaryModel optionally names a smaller/cheaper model for the summary
+	// call. Empty uses the pipeline's own LLM.
+	SummaryModel string `yaml:"summary_model,omitempty" json:"summary_model,omitempty"`
+	// MaxSummaryTokens advises the summary length (fed to the prompt). Default 512.
+	MaxSummaryTokens int `yaml:"max_summary_tokens,omitempty" json:"max_summary_tokens,omitempty"`
+}
+
 // ApplyReasoningEffort resolves the effective reasoning effort — a per-request
 // value (requestEffort) overrides the config's own ReasoningEffort default —
 // stores it on the config so gRPCPredictOpts forwards it to the backend as the

core/config/runtime_settings_persist.go

@@ -5,6 +5,7 @@ import (
 	"errors"
 	"os"
 	"path/filepath"
+	"reflect"
 )
 
 // runtimeSettingsFile is the on-disk filename inside DynamicConfigsDir.
@@ -33,6 +34,35 @@ func (o *ApplicationConfig) ReadPersistedSettings() (RuntimeSettings, error) {
 	return settings, nil
 }
 
+// MergeNonNil overlays every set (non-nil) field of overlay onto the
+// receiver, leaving the receiver's value untouched wherever overlay left a
+// field unset. Every RuntimeSettings field is a pointer precisely so "set"
+// can be told apart from "absent" (see the type doc), which makes this a
+// faithful partial update: a caller that submits only the field it owns
+// changes exactly that field and never clobbers unrelated settings.
+//
+// This is the read-modify-write contract the persistence helpers exist for.
+// UpdateSettingsEndpoint reads the on-disk settings, merges the request body
+// on top, and writes the result — so a focused admin page that POSTs only its
+// own field (the Middleware page sends only mitm_listen; the detector table
+// only pii_default_detectors) no longer nulls every other setting.
+//
+// Reflection keeps the merge total over the struct: a field added to
+// RuntimeSettings later is merged automatically, so the persistence path can
+// never silently drop a new setting the way a hand-maintained field list
+// would. Non-pointer fields (none today) are skipped — they cannot express
+// "absent", so the receiver wins.
+func (s *RuntimeSettings) MergeNonNil(overlay RuntimeSettings) {
+	dst := reflect.ValueOf(s).Elem()
+	src := reflect.ValueOf(overlay)
+	for i := 0; i < src.NumField(); i++ {
+		f := src.Field(i)
+		if f.Kind() == reflect.Pointer && !f.IsNil() {
+			dst.Field(i).Set(f)
+		}
+	}
+}
+
 // WritePersistedSettings serialises the given RuntimeSettings to
 // runtime_settings.json with restricted permissions (it may carry API
 // keys and P2P tokens).

core/config/runtime_settings_persist_test.go

@@ -12,6 +12,7 @@ import (
 )
 
 func strPtr(s string) *string { return &s }
+func boolPtr(b bool) *bool     { return &b }
 
 var _ = Describe("RuntimeSettings persistence helpers", func() {
 	var (
@@ -51,6 +52,47 @@ var _ = Describe("RuntimeSettings persistence helpers", func() {
 		})
 	})
 
+	// MergeNonNil is the partial-update primitive UpdateSettingsEndpoint
+	// relies on: a focused admin page POSTs only the field it owns, and the
+	// handler reads the on-disk settings and overlays the request on top.
+	// Without it, the body would be written verbatim and every field the
+	// caller omitted would be nulled (the reported regression: changing
+	// mitm_listen wiped the galleries, api keys, watchdog config, etc.).
+	Describe("MergeNonNil partial update", func() {
+		It("overlays set fields and preserves unset ones", func() {
+			base := config.RuntimeSettings{
+				MITMListen:          strPtr(":9000"),
+				Galleries:           &[]config.Gallery{{Name: "g1", URL: "http://example/g1"}},
+				WatchdogIdleEnabled: boolPtr(true),
+				ApiKeys:             &[]string{"persisted-key"},
+				PIIDefaultDetectors: &[]string{"det-a"},
+			}
+
+			// Simulate the Middleware proxy tab: only mitm_listen is sent.
+			overlay := config.RuntimeSettings{MITMListen: strPtr(":8443")}
+			base.MergeNonNil(overlay)
+
+			Expect(base.MITMListen).ToNot(BeNil())
+			Expect(*base.MITMListen).To(Equal(":8443"), "set field should be overlaid")
+			// Everything the overlay left unset must survive untouched.
+			Expect(base.Galleries).ToNot(BeNil(), "galleries were clobbered")
+			Expect(*base.Galleries).To(HaveLen(1))
+			Expect(base.WatchdogIdleEnabled).ToNot(BeNil())
+			Expect(*base.WatchdogIdleEnabled).To(BeTrue())
+			Expect(base.ApiKeys).ToNot(BeNil(), "api_keys were clobbered")
+			Expect(*base.ApiKeys).To(Equal([]string{"persisted-key"}))
+			Expect(base.PIIDefaultDetectors).ToNot(BeNil(), "pii_default_detectors were clobbered")
+			Expect(*base.PIIDefaultDetectors).To(Equal([]string{"det-a"}))
+		})
+
+		It("lets an explicit empty slice clear a field", func() {
+			base := config.RuntimeSettings{PIIDefaultDetectors: &[]string{"det-a"}}
+			base.MergeNonNil(config.RuntimeSettings{PIIDefaultDetectors: &[]string{}})
+			Expect(base.PIIDefaultDetectors).ToNot(BeNil())
+			Expect(*base.PIIDefaultDetectors).To(BeEmpty(), "an explicit empty slice should clear, not preserve")
+		})
+	})
+
 	// MITM round trip pins the contract that loadRuntimeSettingsFromFile
 	// MITM listener address must survive a write/read round trip so the
 	// next process restart can bring the listener back up. (Intercept

core/http/endpoints/localai/nodes.go

@@ -385,6 +385,23 @@ func GetNodeModelsEndpoint(registry *nodes.NodeRegistry) echo.HandlerFunc {
 	}
 }
 
+// ListAllNodeModelsEndpoint returns all loaded models across all healthy nodes.
+// @Summary List all loaded models cluster-wide
+// @Tags Nodes
+// @Success 200 {array} nodes.NodeModel
+// @Router /api/nodes/models [get]
+func ListAllNodeModelsEndpoint(registry *nodes.NodeRegistry) echo.HandlerFunc {
+	return func(c echo.Context) error {
+		ctx := c.Request().Context()
+		models, err := registry.ListAllLoadedModels(ctx)
+		if err != nil {
+			xlog.Error("Failed to list all node models", "error", err)
+			return c.JSON(http.StatusInternalServerError, nodeError(http.StatusInternalServerError, "failed to list node models"))
+		}
+		return c.JSON(http.StatusOK, models)
+	}
+}
+
 // DrainNodeEndpoint sets a node to draining status (no new requests).
 func DrainNodeEndpoint(registry *nodes.NodeRegistry) echo.HandlerFunc {
 	return func(c echo.Context) error {

core/http/endpoints/localai/nodes_test.go

@@ -407,4 +407,44 @@ var _ = Describe("Node HTTP handlers", func() {
 			Expect(names).To(ConsistOf("alpha", "beta"))
 		})
 	})
+
+	Describe("ListAllNodeModelsEndpoint", func() {
+		It("returns an empty list when no models are loaded", func() {
+			e := echo.New()
+			req := httptest.NewRequest(http.MethodGet, "/", nil)
+			rec := httptest.NewRecorder()
+			c := e.NewContext(req, rec)
+
+			handler := ListAllNodeModelsEndpoint(registry)
+			Expect(handler(c)).To(Succeed())
+			Expect(rec.Code).To(Equal(http.StatusOK))
+
+			var list []nodes.NodeModel
+			Expect(json.Unmarshal(rec.Body.Bytes(), &list)).To(Succeed())
+			Expect(list).To(BeEmpty())
+		})
+
+		It("returns loaded models across healthy nodes", func() {
+			ctx := context.Background()
+			Expect(registry.Register(ctx, &nodes.BackendNode{
+				ID: "n1", Name: "alpha", Address: "10.0.0.1:50051", Status: nodes.StatusHealthy,
+			}, true)).To(Succeed())
+			Expect(registry.SetNodeModel(ctx, "n1", "llama-3.3", 0, "loaded", "10.0.0.1:50051", 0)).To(Succeed())
+
+			e := echo.New()
+			req := httptest.NewRequest(http.MethodGet, "/", nil)
+			rec := httptest.NewRecorder()
+			c := e.NewContext(req, rec)
+
+			handler := ListAllNodeModelsEndpoint(registry)
+			Expect(handler(c)).To(Succeed())
+			Expect(rec.Code).To(Equal(http.StatusOK))
+
+			var list []nodes.NodeModel
+			Expect(json.Unmarshal(rec.Body.Bytes(), &list)).To(Succeed())
+			Expect(list).To(HaveLen(1))
+			Expect(list[0].ModelName).To(Equal("llama-3.3"))
+			Expect(list[0].NodeID).To(Equal("n1"))
+		})
+	})
 })

core/http/endpoints/localai/settings.go

@@ -4,8 +4,6 @@ import (
 	"encoding/json"
 	"io"
 	"net/http"
-	"os"
-	"path/filepath"
 	"time"
 
 	"github.com/labstack/echo/v4"
@@ -110,6 +108,18 @@ func UpdateSettingsEndpoint(app *application.Application) echo.HandlerFunc {
 			})
 		}
 
+		// Read whatever is already persisted: it is both the source of truth
+		// for branding asset filenames (below) and the base we merge this
+		// request onto before writing. A read failure must not let a Save
+		// silently discard the existing settings — surface it instead.
+		persisted, err := appConfig.ReadPersistedSettings()
+		if err != nil {
+			return c.JSON(http.StatusInternalServerError, schema.SettingsResponse{
+				Success: false,
+				Error:   "Failed to read existing settings: " + err.Error(),
+			})
+		}
+
 		// Branding asset filenames are owned exclusively by
 		// /api/branding/asset/{kind} (upload/delete). The Settings page also
 		// round-trips them via GET /api/settings, but its local state is stale
@@ -118,11 +128,9 @@ func UpdateSettingsEndpoint(app *application.Application) echo.HandlerFunc {
 		// at page open. Replace whatever the body sent for these three fields
 		// with the values currently on disk so /api/settings can never
 		// regress them.
-		if existing, err := appConfig.ReadPersistedSettings(); err == nil {
-			settings.LogoFile = existing.LogoFile
-			settings.LogoHorizontalFile = existing.LogoHorizontalFile
-			settings.FaviconFile = existing.FaviconFile
-		}
+		settings.LogoFile = persisted.LogoFile
+		settings.LogoHorizontalFile = persisted.LogoHorizontalFile
+		settings.FaviconFile = persisted.FaviconFile
 
 		// The UI reads ApiKeys from GET /api/settings, which already returns the
 		// merged env+runtime list. When the user clicks Save, the same merged
@@ -145,16 +153,17 @@ func UpdateSettingsEndpoint(app *application.Application) echo.HandlerFunc {
 			settings.ApiKeys = &runtimeOnly
 		}
 
-		settingsFile := filepath.Join(appConfig.DynamicConfigsDir, "runtime_settings.json")
-		settingsJSON, err := json.MarshalIndent(settings, "", "  ")
-		if err != nil {
-			return c.JSON(http.StatusInternalServerError, schema.SettingsResponse{
-				Success: false,
-				Error:   "Failed to marshal settings: " + err.Error(),
-			})
-		}
-
-		if err := os.WriteFile(settingsFile, settingsJSON, 0600); err != nil {
+		// Persist as a partial update: overlay only the fields this request set
+		// onto the settings already on disk. Focused admin pages POST just the
+		// keys they own (the Middleware proxy tab sends only mitm_listen; the
+		// detector table only pii_default_detectors), so writing the request
+		// body verbatim would null every unrelated setting (the no-omitempty
+		// api_keys / pii_default_detectors fields even round-trip as JSON
+		// null). The full Settings page still round-trips every field, so its
+		// Save is unchanged.
+		toPersist := persisted
+		toPersist.MergeNonNil(settings)
+		if err := appConfig.WritePersistedSettings(toPersist); err != nil {
 			return c.JSON(http.StatusInternalServerError, schema.SettingsResponse{
 				Success: false,
 				Error:   "Failed to write settings file: " + err.Error(),

core/http/endpoints/localai/settings_test.go

@@ -52,6 +52,10 @@ var _ = Describe("Settings endpoints", func() {
 		// Settings are persisted here; set after construction since there's no
 		// dedicated AppOption for it.
 		app.ApplicationConfig().DynamicConfigsDir = tmp
+		// Contain the MITM CA inside tmp too. The partial-save spec flips
+		// mitm_listen, which starts the listener and writes a CA; without this
+		// it defaults to ./mitm-ca and litters the package source tree.
+		app.ApplicationConfig().MITMCADir = filepath.Join(tmp, "mitm-ca")
 
 		e = echo.New()
 		e.GET("/api/settings", GetSettingsEndpoint(app))
@@ -109,6 +113,39 @@ var _ = Describe("Settings endpoints", func() {
 		Expect(err).ToNot(HaveOccurred())
 	})
 
+	// Regression: a focused admin page (the Middleware proxy tab) POSTs only
+	// the one field it owns — mitm_listen. The old handler wrote the request
+	// body verbatim, so every other persisted setting was dropped (and
+	// api_keys / pii_default_detectors, which lack omitempty, were written as
+	// null). A partial POST must now merge onto what is already on disk.
+	It("preserves unrelated persisted settings when a partial POST sets only mitm_listen", func() {
+		// First save establishes a fuller settings file (as the full Settings
+		// page would): galleries, an API key, and the MITM listener. The
+		// listener restart binds a real socket, so use 127.0.0.1:0 for an
+		// ephemeral free port rather than a fixed one that may be in use.
+		rec := post(`{"mitm_listen":"127.0.0.1:0","galleries":[{"name":"g1","url":"http://example/g1"}],"api_keys":["k1"],"pii_default_detectors":["det-a"]}`)
+		Expect(rec.Code).To(Equal(http.StatusOK), rec.Body.String())
+
+		// The Middleware proxy tab then changes only the listen address — the
+		// exact partial body that nulled everything else before the fix.
+		rec = post(`{"mitm_listen":"127.0.0.1:0"}`)
+		Expect(rec.Code).To(Equal(http.StatusOK), rec.Body.String())
+
+		raw, err := os.ReadFile(filepath.Join(tmp, "runtime_settings.json"))
+		Expect(err).ToNot(HaveOccurred())
+		var ondisk config.RuntimeSettings
+		Expect(json.Unmarshal(raw, &ondisk)).To(Succeed())
+
+		Expect(ondisk.MITMListen).ToNot(BeNil())
+		Expect(*ondisk.MITMListen).To(Equal("127.0.0.1:0"), "the changed field should be saved")
+		Expect(ondisk.Galleries).ToNot(BeNil(), "galleries were clobbered by the partial save")
+		Expect(*ondisk.Galleries).To(HaveLen(1))
+		Expect(ondisk.ApiKeys).ToNot(BeNil(), "api_keys were nulled by the partial save")
+		Expect(*ondisk.ApiKeys).To(Equal([]string{"k1"}))
+		Expect(ondisk.PIIDefaultDetectors).ToNot(BeNil(), "pii_default_detectors were nulled by the partial save")
+		Expect(*ondisk.PIIDefaultDetectors).To(Equal([]string{"det-a"}))
+	})
+
 	// Residual #9125: enabling the watchdog from a cold (off) state via the
 	// React master toggle must start the live watchdog immediately, without a
 	// restart. The toggle posts watchdog_idle_enabled/busy_enabled=true while

core/http/endpoints/openai/realtime.go

@@ -12,6 +12,7 @@ import (
 	"os"
 	"strconv"
 	"sync"
+	"sync/atomic"
 	"time"
 
 	"net/http"
@@ -134,6 +135,18 @@ type Session struct {
 	// pairs are kept together so we never feed an orphaned tool result.
 	MaxHistoryItems int
 
+	// Compaction settings resolved from pipeline.compaction (see resolveCompaction).
+	CompactionEnabled bool
+	CompactionTrigger int
+	SummaryModel      string
+	MaxSummaryTokens  int
+
+	// summarizerFactory lazily builds the model used for compaction summaries
+	// when summary_model is configured; nil means reuse the pipeline LLM.
+	summarizerFactory func() (Model, error)
+	summarizerOnce    sync.Once
+	summarizerCached  Model
+
 	// AssistantExecutor is non-nil when the session opted into the in-process
 	// LocalAI Assistant tool surface. Tool calls whose name matches this
 	// executor's catalog are run inproc and their output is fed back to the
@@ -241,6 +254,12 @@ type Conversation struct {
 	ID    string
 	Items []*types.MessageItemUnion
 	Lock  sync.Mutex
+	// Memory is the rolling summary of items already evicted by compaction. It
+	// is kept out of Items (so trimRealtimeItems never drops it) and rendered
+	// as a system message right after the session instructions.
+	Memory string
+	// compacting ensures at most one background compaction runs per conversation.
+	compacting atomic.Bool
 }
 
 func (c *Conversation) ToServer() types.Conversation {
@@ -540,13 +559,12 @@ func runRealtimeSession(application *application.Application, t Transport, model
 		SoundDetectionWindowMs:  cfg.Pipeline.SoundDetectionWindowMs,
 		SoundDetectionHopMs:     cfg.Pipeline.SoundDetectionHopMs,
 	}
+	session.CompactionEnabled, session.CompactionTrigger, session.MaxSummaryTokens, session.SummaryModel = resolveCompaction(cfg, session.MaxHistoryItems)
 
 	// Create a default conversation
 	conversationID := generateConversationID()
 	conversation := &Conversation{
-		ID: conversationID,
-		// TODO: We need to truncate the conversation items when a new item is added and we have run out of space. There are multiple places where items
-		//       can be added so we could use a datastructure here that enforces truncation upon addition
+		ID:    conversationID,
 		Items: []*types.MessageItemUnion{},
 	}
 	session.Conversations[conversationID] = conversation
@@ -577,6 +595,18 @@ func runRealtimeSession(application *application.Application, t Transport, model
 	}
 	session.ModelInterface = m
 
+	if session.SummaryModel != "" {
+		summaryModelName := session.SummaryModel
+		sid := sessionID
+		session.summarizerFactory = func() (Model, error) {
+			summaryCfg, lerr := application.ModelConfigLoader().LoadModelConfigFileByNameDefaultOptions(summaryModelName, application.ApplicationConfig())
+			if lerr != nil {
+				return nil, fmt.Errorf("load summary model config %q: %w", summaryModelName, lerr)
+			}
+			return newModel(&summaryCfg.Pipeline, application.ModelConfigLoader(), application.ModelLoader(), application.ApplicationConfig(), evaluator, buildRealtimeRoutingContext(application, sid))
+		}
+	}
+
 	if cfg.Pipeline.VoiceGateEnabled() {
 		gate, gerr := newVoiceGate(
 			*cfg.Pipeline.VoiceRecognition,
@@ -807,6 +837,15 @@ func runRealtimeSession(application *application.Application, t Transport, model
 				commitUtterance(respCtx, allAudio, session, conversation, t)
 			}()
 
+		case types.InputAudioBufferClearEvent:
+			xlog.Debug("recv", "message", string(msg))
+			// Discard a partially-captured utterance so the client can restart
+			// input cleanly without the stale buffer leaking into the next commit.
+			clearInputAudio(session)
+			sendEvent(t, types.InputAudioBufferClearedEvent{
+				ServerEventBase: types.ServerEventBase{EventID: e.EventID},
+			})
+
 		case types.ConversationItemCreateEvent:
 			xlog.Debug("recv", "message", string(msg))
 			// Add the item to the conversation
@@ -841,7 +880,39 @@ func runRealtimeSession(application *application.Application, t Transport, model
 			})
 
 		case types.ConversationItemDeleteEvent:
-			sendError(t, "not_implemented", "Deleting items not implemented", "", "event_TODO")
+			xlog.Debug("recv", "message", string(msg))
+			if e.ItemID == "" {
+				sendError(t, "invalid_item_id", "Need item_id, but none specified", "", "event_TODO")
+				continue
+			}
+			conversation.Lock.Lock()
+			updated, ok := deleteItem(conversation.Items, e.ItemID)
+			conversation.Items = updated
+			conversation.Lock.Unlock()
+			if !ok {
+				sendError(t, "invalid_item_id", "Item to delete not found", "", "event_TODO")
+				continue
+			}
+			sendEvent(t, types.ConversationItemDeletedEvent{
+				ServerEventBase: types.ServerEventBase{EventID: e.EventID},
+				ItemID:          e.ItemID,
+			})
+
+		case types.ConversationItemTruncateEvent:
+			xlog.Debug("recv", "message", string(msg))
+			conversation.Lock.Lock()
+			ok := truncateAssistantText(conversation.Items, e.ItemID, e.ContentIndex)
+			conversation.Lock.Unlock()
+			if !ok {
+				sendError(t, "invalid_item_id", "Item to truncate not found", "", "event_TODO")
+				continue
+			}
+			sendEvent(t, types.ConversationItemTruncatedEvent{
+				ServerEventBase: types.ServerEventBase{EventID: e.EventID},
+				ItemID:          e.ItemID,
+				ContentIndex:    e.ContentIndex,
+				AudioEndMs:      e.AudioEndMs,
+			})
 
 		case types.ConversationItemRetrieveEvent:
 			xlog.Debug("recv", "message", string(msg))
@@ -854,21 +925,7 @@ func runRealtimeSession(application *application.Application, t Transport, model
 			conversation.Lock.Lock()
 			var retrievedItem types.MessageItemUnion
 			for _, item := range conversation.Items {
-				// We need to check ID in the union
-				var id string
-				if item.System != nil {
-					id = item.System.ID
-				} else if item.User != nil {
-					id = item.User.ID
-				} else if item.Assistant != nil {
-					id = item.Assistant.ID
-				} else if item.FunctionCall != nil {
-					id = item.FunctionCall.ID
-				} else if item.FunctionCallOutput != nil {
-					id = item.FunctionCallOutput.ID
-				}
-
-				if id == e.ItemID {
+				if itemID(item) == e.ItemID {
 					retrievedItem = *item
 					break
 				}
@@ -1666,6 +1723,9 @@ const maxAssistantToolTurns = 10
 
 func triggerResponse(ctx context.Context, session *Session, conv *Conversation, t Transport, overrides *types.ResponseCreateParams) {
 	triggerResponseAtTurn(ctx, session, conv, t, overrides, 0)
+	// Fold aged-out turns into the rolling memory off the critical path; the
+	// next turn reaps the smaller buffer.
+	session.maybeCompact(conv)
 }
 
 func triggerResponseAtTurn(ctx context.Context, session *Session, conv *Conversation, t Transport, overrides *types.ResponseCreateParams, toolTurn int) {
@@ -1721,6 +1781,7 @@ func triggerResponseAtTurn(ctx context.Context, session *Session, conv *Conversa
 	var lastUserSpeaker *types.Speaker
 	personalize := session.voiceGate != nil && session.voiceGate.cfg.PersonalizeEnabled()
 	conv.Lock.Lock()
+	conversationHistory = withMemory(conversationHistory, conv.Memory)
 	items := trimRealtimeItems(conv.Items, session.MaxHistoryItems)
 	for _, item := range items {
 		if item.User != nil {

core/http/endpoints/openai/realtime_compaction.go

@@ -0,0 +1,326 @@
+package openai
+
+import (
+	"context"
+	"fmt"
+	"strings"
+	"time"
+
+	"github.com/mudler/LocalAI/core/config"
+	"github.com/mudler/LocalAI/core/http/endpoints/openai/types"
+	"github.com/mudler/LocalAI/core/schema"
+	"github.com/mudler/LocalAI/pkg/reasoning"
+	"github.com/mudler/xlog"
+)
+
+const (
+	defaultMaxSummaryTokens = 512
+	memoryPrefix            = "Summary of earlier conversation:\n"
+	// compactionTimeout bounds the summarizer call so a stuck model can't pin the
+	// compacting flag (and thus block all further compaction) forever.
+	compactionTimeout = 60 * time.Second
+)
+
+// withMemory inserts the rolling summary as a system message after the existing
+// (instructions) history. No-op when memory is empty.
+func withMemory(history schema.Messages, memory string) schema.Messages {
+	if memory == "" {
+		return history
+	}
+	content := memoryPrefix + memory
+	return append(history, schema.Message{
+		Role:          string(types.MessageRoleSystem),
+		StringContent: content,
+		Content:       content,
+	})
+}
+
+// renderItemsTranscript renders conversation items as a plain "role: text"
+// transcript for summarization. Non-text items (bare tool calls) are labelled
+// so the summarizer keeps track of actions taken.
+func renderItemsTranscript(items []*types.MessageItemUnion) string {
+	var b strings.Builder
+	for _, item := range items {
+		switch {
+		case item.User != nil:
+			b.WriteString("user: ")
+			for _, c := range item.User.Content {
+				if c.Text != "" {
+					b.WriteString(c.Text)
+				}
+				if c.Transcript != "" {
+					b.WriteString(c.Transcript)
+				}
+			}
+			b.WriteString("\n")
+		case item.Assistant != nil:
+			b.WriteString("assistant: ")
+			// Realtime assistant *audio* turns store the spoken words in
+			// .Transcript (not .Text), so emit both or spoken turns are dropped.
+			for _, c := range item.Assistant.Content {
+				if c.Text != "" {
+					b.WriteString(c.Text)
+				}
+				if c.Transcript != "" {
+					b.WriteString(c.Transcript)
+				}
+			}
+			b.WriteString("\n")
+		case item.FunctionCall != nil:
+			b.WriteString(fmt.Sprintf("assistant called tool %s(%s)\n", item.FunctionCall.Name, item.FunctionCall.Arguments))
+		case item.FunctionCallOutput != nil:
+			b.WriteString(fmt.Sprintf("tool result: %s\n", item.FunctionCallOutput.Output))
+		}
+	}
+	return strings.TrimSpace(b.String())
+}
+
+// buildSummaryMessages builds the chat messages for the summarizer LLM: a system
+// instruction plus prior memory and the new transcript to fold in. maxTokens is
+// advisory (fed to the prompt; not hard-enforced in v1).
+func buildSummaryMessages(priorMemory, transcript string, maxTokens int) schema.Messages {
+	system := fmt.Sprintf("You maintain a running memory of a live voice conversation. "+
+		"Merge the prior memory with the new exchanges into an updated memory. "+
+		"Keep names, decisions, facts, preferences, and open threads. Be concise "+
+		"(under ~%d tokens). Output only the updated memory, with no reasoning or tags.", maxTokens)
+	var user strings.Builder
+	if priorMemory != "" {
+		user.WriteString("Prior memory:\n")
+		user.WriteString(priorMemory)
+		user.WriteString("\n\n")
+	}
+	user.WriteString("New exchanges to fold in:\n")
+	user.WriteString(transcript)
+	return schema.Messages{
+		{Role: string(types.MessageRoleSystem), StringContent: system, Content: system},
+		{Role: string(types.MessageRoleUser), StringContent: user.String(), Content: user.String()},
+	}
+}
+
+// clearInputAudio resets the session's pending input audio buffer (the raw
+// PCM and any buffered Opus frames). Used by the input_audio_buffer.clear
+// realtime event so a client can discard a partially-captured utterance.
+func clearInputAudio(s *Session) {
+	s.AudioBufferLock.Lock()
+	s.InputAudioBuffer = nil
+	s.AudioBufferLock.Unlock()
+	s.OpusFramesLock.Lock()
+	s.OpusFrames = nil
+	s.OpusFramesLock.Unlock()
+}
+
+// itemID extracts the id from any MessageItemUnion variant ("" if none).
+func itemID(item *types.MessageItemUnion) string {
+	switch {
+	case item == nil:
+		return ""
+	case item.System != nil:
+		return item.System.ID
+	case item.User != nil:
+		return item.User.ID
+	case item.Assistant != nil:
+		return item.Assistant.ID
+	case item.FunctionCall != nil:
+		return item.FunctionCall.ID
+	case item.FunctionCallOutput != nil:
+		return item.FunctionCallOutput.ID
+	default:
+		return ""
+	}
+}
+
+// deleteItem removes the item with id from items, returning the new slice and
+// whether it was found.
+func deleteItem(items []*types.MessageItemUnion, id string) ([]*types.MessageItemUnion, bool) {
+	for i, item := range items {
+		if itemID(item) == id {
+			return append(items[:i:i], items[i+1:]...), true
+		}
+	}
+	return items, false
+}
+
+// truncateAssistantText clears the text of the assistant item's content part at
+// contentIndex. Minimal truncate: used to discard an interrupted/barge-in
+// response tail. Both .Text and .Transcript are cleared because realtime audio
+// turns store the spoken words in .Transcript (clearing only .Text would no-op).
+func truncateAssistantText(items []*types.MessageItemUnion, id string, contentIndex int) bool {
+	for _, item := range items {
+		if itemID(item) != id || item.Assistant == nil {
+			continue
+		}
+		if contentIndex >= 0 && contentIndex < len(item.Assistant.Content) {
+			item.Assistant.Content[contentIndex].Text = ""
+			item.Assistant.Content[contentIndex].Transcript = ""
+		}
+		return true
+	}
+	return false
+}
+
+// compactionCut returns the index splitting items into overflow (items[:cut],
+// to be summarized+evicted) and the kept live tail (items[cut:]), keeping the
+// last `keep` items. It mirrors trimRealtimeItems' pair-safety: the cut is
+// pulled left so a function_call and its function_call_output are never split
+// across the boundary (the whole pair lands in the kept tail). Returns 0 when
+// there is nothing to cut.
+func compactionCut(items []*types.MessageItemUnion, keep int) int {
+	// keep <= 0 means no live-window cap (the "unlimited history" sentinel, as
+	// in trimRealtimeItems): there is nothing to evict, so cut nothing. This
+	// also avoids indexing items[len(items)] in the pair-safety loop below.
+	if keep <= 0 {
+		return 0
+	}
+	cut := len(items) - keep
+	if cut <= 0 {
+		return 0
+	}
+	for cut > 0 && items[cut] != nil && items[cut].FunctionCallOutput != nil {
+		cut--
+	}
+	return cut
+}
+
+// resolveCompaction reads the pipeline.compaction block, applying defaults and
+// the trigger>max_history invariant. maxHistory is the already-resolved live
+// window size. Returns enabled=false (and zero values) when compaction is off.
+func resolveCompaction(cfg *config.ModelConfig, maxHistory int) (enabled bool, trigger, maxSummaryTokens int, summaryModel string) {
+	if cfg == nil || cfg.Pipeline.Compaction == nil || !cfg.Pipeline.Compaction.Enabled {
+		return false, 0, 0, ""
+	}
+	c := cfg.Pipeline.Compaction
+	trigger = c.TriggerItems
+	if trigger <= 0 {
+		trigger = maxHistory * 2
+	}
+	if trigger <= maxHistory {
+		trigger = maxHistory + 1
+	}
+	maxSummaryTokens = c.MaxSummaryTokens
+	if maxSummaryTokens <= 0 {
+		maxSummaryTokens = defaultMaxSummaryTokens
+	}
+	return true, trigger, maxSummaryTokens, c.SummaryModel
+}
+
+// prefixMatches reports whether items begins with the same ids, in order, as
+// snapshot — i.e. the overflow we summarized is still at the head (no concurrent
+// client delete reshuffled it).
+func prefixMatches(items, snapshot []*types.MessageItemUnion) bool {
+	if len(items) < len(snapshot) {
+		return false
+	}
+	for i := range snapshot {
+		if itemID(items[i]) != itemID(snapshot[i]) {
+			return false
+		}
+	}
+	return true
+}
+
+// compact folds overflow items into conv.Memory and evicts them. It never holds
+// conv.Lock across the summarizer call: snapshot under lock, summarize unlocked,
+// commit under lock (re-validating the head is unchanged). On any error it
+// leaves the conversation untouched — items are never dropped without a summary.
+func (s *Session) compact(conv *Conversation, model Model) {
+	if model == nil {
+		return
+	}
+	// Snapshot.
+	conv.Lock.Lock()
+	if len(conv.Items) <= s.CompactionTrigger {
+		conv.Lock.Unlock()
+		return
+	}
+	cut := compactionCut(conv.Items, s.MaxHistoryItems)
+	if cut <= 0 {
+		conv.Lock.Unlock()
+		return
+	}
+	overflow := append([]*types.MessageItemUnion(nil), conv.Items[:cut]...)
+	prior := conv.Memory
+	conv.Lock.Unlock()
+
+	// Summarize (unlocked).
+	msgs := buildSummaryMessages(prior, renderItemsTranscript(overflow), s.MaxSummaryTokens)
+	ctx, cancel := context.WithTimeout(context.Background(), compactionTimeout)
+	defer cancel()
+	predFunc, err := model.Predict(ctx, msgs, nil, nil, nil, nil, nil, nil, nil, nil, nil)
+	if err != nil {
+		xlog.Warn("realtime compaction: summarizer predict failed", "error", err)
+		return
+	}
+	pred, err := predFunc()
+	if err != nil {
+		xlog.Warn("realtime compaction: summarizer inference failed", "error", err)
+		return
+	}
+	// Strip any leaked reasoning/thinking spans using the same extractor the
+	// rest of the realtime path uses, rather than a bespoke regex.
+	rcfg := reasoning.Config{}
+	if mc := model.PredictConfig(); mc != nil {
+		rcfg = spokenReasoningConfig(mc.ReasoningConfig)
+	}
+	_, summary := reasoning.ExtractReasoningComplete(pred.Response, "", rcfg)
+	summary = strings.TrimSpace(summary)
+	if summary == "" {
+		xlog.Warn("realtime compaction: empty summary, skipping eviction")
+		return
+	}
+
+	// Commit.
+	conv.Lock.Lock()
+	defer conv.Lock.Unlock()
+	if !prefixMatches(conv.Items, overflow) {
+		xlog.Debug("realtime compaction: head changed during summary, skipping")
+		return
+	}
+	conv.Memory = summary
+	conv.Items = conv.Items[len(overflow):]
+	xlog.Debug("realtime compaction: evicted items into memory", "evicted", len(overflow), "remaining", len(conv.Items))
+}
+
+// summarizerModel resolves the model used to produce compaction summaries.
+// Without a configured summary_model (or factory) it reuses the pipeline LLM.
+func (s *Session) summarizerModel() Model {
+	if s.SummaryModel == "" || s.summarizerFactory == nil {
+		return s.ModelInterface
+	}
+	s.summarizerOnce.Do(func() {
+		m, err := s.summarizerFactory()
+		if err != nil {
+			xlog.Warn("realtime compaction: summary_model load failed, falling back to pipeline LLM", "model", s.SummaryModel, "error", err)
+			m = s.ModelInterface
+		}
+		s.summarizerCached = m
+	})
+	return s.summarizerCached
+}
+
+// maybeCompact schedules a background compaction when the live buffer has grown
+// past the trigger and none is already running. Returns immediately.
+func (s *Session) maybeCompact(conv *Conversation) {
+	if !s.CompactionEnabled {
+		return
+	}
+	conv.Lock.Lock()
+	over := len(conv.Items) > s.CompactionTrigger
+	conv.Lock.Unlock()
+	if !over {
+		return
+	}
+	if !conv.compacting.CompareAndSwap(false, true) {
+		return
+	}
+	go func() {
+		defer conv.compacting.Store(false)
+		// Resolve (and, for a configured summary_model, lazily load) the
+		// summarizer only when a compaction actually runs, off the response
+		// path — so the model load never blocks a user turn.
+		model := s.summarizerModel()
+		if model == nil {
+			return
+		}
+		s.compact(conv, model)
+	}()
+}

core/http/endpoints/openai/realtime_compaction_test.go

@@ -0,0 +1,308 @@
+package openai
+
+import (
+	"errors"
+
+	. "github.com/onsi/ginkgo/v2"
+	. "github.com/onsi/gomega"
+
+	"github.com/mudler/LocalAI/core/backend"
+	"github.com/mudler/LocalAI/core/config"
+	"github.com/mudler/LocalAI/core/http/endpoints/openai/types"
+	"github.com/mudler/LocalAI/core/schema"
+)
+
+var _ = Describe("resolveCompaction", func() {
+	It("disables when the block is absent", func() {
+		enabled, _, _, _ := resolveCompaction(&config.ModelConfig{}, 6)
+		Expect(enabled).To(BeFalse())
+	})
+
+	It("defaults trigger to 2x max history and tokens to 512", func() {
+		cfg := &config.ModelConfig{Pipeline: config.Pipeline{Compaction: &config.PipelineCompaction{Enabled: true}}}
+		enabled, trigger, maxTok, _ := resolveCompaction(cfg, 6)
+		Expect(enabled).To(BeTrue())
+		Expect(trigger).To(Equal(12))
+		Expect(maxTok).To(Equal(512))
+	})
+
+	It("clamps trigger to max history + 1 when misconfigured", func() {
+		cfg := &config.ModelConfig{Pipeline: config.Pipeline{Compaction: &config.PipelineCompaction{Enabled: true, TriggerItems: 4}}}
+		_, trigger, _, _ := resolveCompaction(cfg, 6)
+		Expect(trigger).To(Equal(7))
+	})
+
+	It("honors explicit values", func() {
+		cfg := &config.ModelConfig{Pipeline: config.Pipeline{Compaction: &config.PipelineCompaction{
+			Enabled: true, TriggerItems: 20, MaxSummaryTokens: 256, SummaryModel: "tiny"}}}
+		enabled, trigger, maxTok, model := resolveCompaction(cfg, 6)
+		Expect(enabled).To(BeTrue())
+		Expect(trigger).To(Equal(20))
+		Expect(maxTok).To(Equal(256))
+		Expect(model).To(Equal("tiny"))
+	})
+})
+
+var _ = Describe("deleteItem", func() {
+	mk := func(ids ...string) []*types.MessageItemUnion {
+		out := make([]*types.MessageItemUnion, len(ids))
+		for i, id := range ids {
+			out[i] = &types.MessageItemUnion{User: &types.MessageItemUser{ID: id}}
+		}
+		return out
+	}
+
+	It("removes the item with the given id", func() {
+		items, ok := deleteItem(mk("a", "b", "c"), "b")
+		Expect(ok).To(BeTrue())
+		Expect(len(items)).To(Equal(2))
+		Expect(itemID(items[0])).To(Equal("a"))
+		Expect(itemID(items[1])).To(Equal("c"))
+	})
+
+	It("reports not found for an unknown id", func() {
+		_, ok := deleteItem(mk("a"), "zzz")
+		Expect(ok).To(BeFalse())
+	})
+})
+
+var _ = Describe("clearInputAudio", func() {
+	It("resets the pending PCM and buffered Opus frames", func() {
+		s := &Session{InputAudioBuffer: []byte{1, 2, 3}, OpusFrames: [][]byte{{9}}}
+		clearInputAudio(s)
+		Expect(s.InputAudioBuffer).To(BeNil())
+		Expect(s.OpusFrames).To(BeNil())
+	})
+})
+
+var _ = Describe("truncateAssistantText", func() {
+	It("clears the text of the assistant content part at the index", func() {
+		items := []*types.MessageItemUnion{{Assistant: &types.MessageItemAssistant{
+			ID:      "a1",
+			Content: []types.MessageContentOutput{{Type: types.MessageContentTypeText, Text: "hello world"}},
+		}}}
+		ok := truncateAssistantText(items, "a1", 0)
+		Expect(ok).To(BeTrue())
+		Expect(items[0].Assistant.Content[0].Text).To(Equal(""))
+	})
+
+	// Realtime assistant *audio* turns store the spoken words in .Transcript, not
+	// .Text, so a barge-in truncate must clear .Transcript too or it would no-op.
+	It("clears the transcript of an assistant audio content part", func() {
+		items := []*types.MessageItemUnion{{Assistant: &types.MessageItemAssistant{
+			ID:      "a1",
+			Content: []types.MessageContentOutput{{Type: types.MessageContentTypeAudio, Transcript: "hello world"}},
+		}}}
+		ok := truncateAssistantText(items, "a1", 0)
+		Expect(ok).To(BeTrue())
+		Expect(items[0].Assistant.Content[0].Transcript).To(Equal(""))
+	})
+
+	It("returns false for an unknown id", func() {
+		Expect(truncateAssistantText(nil, "nope", 0)).To(BeFalse())
+	})
+})
+
+var _ = Describe("compactionCut", func() {
+	user := func(id string) *types.MessageItemUnion {
+		return &types.MessageItemUnion{User: &types.MessageItemUser{ID: id}}
+	}
+	call := func(id string) *types.MessageItemUnion {
+		return &types.MessageItemUnion{FunctionCall: &types.MessageItemFunctionCall{ID: id}}
+	}
+	out := func(id string) *types.MessageItemUnion {
+		return &types.MessageItemUnion{FunctionCallOutput: &types.MessageItemFunctionCallOutput{ID: id}}
+	}
+
+	It("cuts exactly len-keep when no pairs straddle the boundary", func() {
+		items := []*types.MessageItemUnion{user("1"), user("2"), user("3"), user("4")}
+		Expect(compactionCut(items, 2)).To(Equal(2))
+	})
+
+	It("returns 0 when nothing to cut", func() {
+		Expect(compactionCut([]*types.MessageItemUnion{user("1")}, 2)).To(Equal(0))
+	})
+
+	It("returns 0 (cuts nothing) when keep is 0 — the unlimited-window sentinel", func() {
+		items := []*types.MessageItemUnion{user("1"), user("2"), user("3")}
+		Expect(compactionCut(items, 0)).To(Equal(0))
+	})
+
+	It("moves the boundary so a call/output pair is not split", func() {
+		// keep=2 -> naive cut=2, but items[2] is the output of items[1]'s call;
+		// pull the cut right so the whole pair stays in the kept tail.
+		items := []*types.MessageItemUnion{user("1"), call("c"), out("c"), user("4")}
+		Expect(compactionCut(items, 2)).To(Equal(1))
+	})
+})
+
+var _ = Describe("withMemory", func() {
+	It("inserts a memory system message when memory is non-empty", func() {
+		base := schema.Messages{{Role: "system", StringContent: "instructions"}}
+		out := withMemory(base, "user is Bob; wants pizza")
+		Expect(len(out)).To(Equal(2))
+		Expect(out[1].Role).To(Equal("system"))
+		Expect(out[1].StringContent).To(ContainSubstring("user is Bob"))
+		Expect(out[1].StringContent).To(ContainSubstring("Summary of earlier conversation"))
+	})
+
+	It("is a no-op when memory is empty", func() {
+		base := schema.Messages{{Role: "system", StringContent: "instructions"}}
+		Expect(withMemory(base, "")).To(HaveLen(1))
+	})
+})
+
+var _ = Describe("renderItemsTranscript", func() {
+	It("renders user and assistant text turns", func() {
+		items := []*types.MessageItemUnion{
+			{User: &types.MessageItemUser{Content: []types.MessageContentInput{{Type: types.MessageContentTypeInputText, Text: "hi"}}}},
+			{Assistant: &types.MessageItemAssistant{Content: []types.MessageContentOutput{{Type: types.MessageContentTypeText, Text: "hello"}}}},
+		}
+		out := renderItemsTranscript(items)
+		Expect(out).To(ContainSubstring("user: hi"))
+		Expect(out).To(ContainSubstring("assistant: hello"))
+	})
+
+	// Realtime assistant *audio* turns store the spoken words in .Transcript, not
+	// .Text, so the transcript builder must emit .Transcript too or spoken turns
+	// would be dropped from the summary.
+	It("renders an assistant audio turn from its transcript", func() {
+		items := []*types.MessageItemUnion{
+			{Assistant: &types.MessageItemAssistant{Content: []types.MessageContentOutput{{Type: types.MessageContentTypeAudio, Transcript: "spoken words"}}}},
+		}
+		Expect(renderItemsTranscript(items)).To(ContainSubstring("assistant: spoken words"))
+	})
+})
+
+var _ = Describe("buildSummaryMessages", func() {
+	It("includes prior memory and the new transcript", func() {
+		msgs := buildSummaryMessages("prior facts", "user: hi", 512)
+		Expect(len(msgs)).To(Equal(2))
+		Expect(msgs[0].Role).To(Equal("system"))
+		Expect(msgs[1].StringContent).To(ContainSubstring("prior facts"))
+		Expect(msgs[1].StringContent).To(ContainSubstring("user: hi"))
+	})
+})
+
+var _ = Describe("compact", func() {
+	user := func(id, text string) *types.MessageItemUnion {
+		return &types.MessageItemUnion{User: &types.MessageItemUser{ID: id,
+			Content: []types.MessageContentInput{{Type: types.MessageContentTypeInputText, Text: text}}}}
+	}
+
+	It("summarizes overflow into Memory and evicts it, keeping the live tail", func() {
+		conv := &Conversation{Items: []*types.MessageItemUnion{
+			user("1", "a"), user("2", "b"), user("3", "c"), user("4", "d"),
+			user("5", "e"), user("6", "f"), user("7", "g"), user("8", "h"),
+		}}
+		s := &Session{CompactionEnabled: true, CompactionTrigger: 7, MaxHistoryItems: 4, MaxSummaryTokens: 512}
+		m := &fakeModel{predictResp: backend.LLMResponse{Response: "ROLLED UP"}}
+
+		s.compact(conv, m)
+
+		Expect(conv.Memory).To(Equal("ROLLED UP"))
+		Expect(len(conv.Items)).To(Equal(4))
+		Expect(itemID(conv.Items[0])).To(Equal("5"))
+		// The summarizer saw the evicted turns.
+		Expect(m.lastMessages[1].StringContent).To(ContainSubstring("a"))
+	})
+
+	It("leaves Items and Memory untouched when the summarizer errors", func() {
+		items := []*types.MessageItemUnion{user("1", "a"), user("2", "b"), user("3", "c")}
+		conv := &Conversation{Items: items}
+		s := &Session{CompactionEnabled: true, CompactionTrigger: 2, MaxHistoryItems: 1, MaxSummaryTokens: 512}
+		m := &fakeModel{predictErr: errors.New("boom")}
+
+		s.compact(conv, m)
+
+		Expect(conv.Memory).To(Equal(""))
+		Expect(len(conv.Items)).To(Equal(3))
+	})
+
+	It("strips leaked reasoning tags from the summary via the shared extractor", func() {
+		conv := &Conversation{Items: []*types.MessageItemUnion{
+			user("1", "a"), user("2", "b"), user("3", "c"), user("4", "d"),
+			user("5", "e"), user("6", "f"), user("7", "g"), user("8", "h"),
+		}}
+		s := &Session{CompactionEnabled: true, CompactionTrigger: 7, MaxHistoryItems: 4, MaxSummaryTokens: 512}
+		m := &fakeModel{predictResp: backend.LLMResponse{Response: "<think>planning the summary</think>CLEAN SUMMARY"}}
+
+		s.compact(conv, m)
+
+		Expect(conv.Memory).To(Equal("CLEAN SUMMARY"))
+		Expect(conv.Memory).ToNot(ContainSubstring("planning"))
+	})
+
+	It("does nothing when items are at or below the trigger", func() {
+		conv := &Conversation{Items: []*types.MessageItemUnion{user("1", "a")}}
+		s := &Session{CompactionEnabled: true, CompactionTrigger: 7, MaxHistoryItems: 4}
+		s.compact(conv, &fakeModel{predictResp: backend.LLMResponse{Response: "x"}})
+		Expect(conv.Memory).To(Equal(""))
+		Expect(len(conv.Items)).To(Equal(1))
+	})
+})
+
+var _ = Describe("prefixMatches", func() {
+	user := func(id string) *types.MessageItemUnion {
+		return &types.MessageItemUnion{User: &types.MessageItemUser{ID: id}}
+	}
+
+	It("matches when items begins with the snapshot ids in order", func() {
+		items := []*types.MessageItemUnion{user("1"), user("2"), user("3")}
+		snap := []*types.MessageItemUnion{user("1"), user("2")}
+		Expect(prefixMatches(items, snap)).To(BeTrue())
+	})
+
+	It("matches an empty snapshot", func() {
+		Expect(prefixMatches([]*types.MessageItemUnion{user("1")}, nil)).To(BeTrue())
+	})
+
+	It("fails when items is shorter than the snapshot (a concurrent delete shrank the head)", func() {
+		items := []*types.MessageItemUnion{user("1")}
+		snap := []*types.MessageItemUnion{user("1"), user("2")}
+		Expect(prefixMatches(items, snap)).To(BeFalse())
+	})
+
+	It("fails when the head ids differ (a concurrent delete reordered the head)", func() {
+		items := []*types.MessageItemUnion{user("2"), user("3")}
+		snap := []*types.MessageItemUnion{user("1"), user("2")}
+		Expect(prefixMatches(items, snap)).To(BeFalse())
+	})
+})
+
+var _ = Describe("summarizerModel", func() {
+	It("returns the pipeline model when no summary_model is set", func() {
+		m := &fakeModel{}
+		s := &Session{ModelInterface: m}
+		Expect(s.summarizerModel()).To(Equal(m))
+	})
+
+	It("uses the factory (once) when summary_model is set", func() {
+		pipeline := &fakeModel{}
+		small := &fakeModel{}
+		calls := 0
+		s := &Session{ModelInterface: pipeline, SummaryModel: "tiny",
+			summarizerFactory: func() (Model, error) { calls++; return small, nil }}
+		Expect(s.summarizerModel()).To(Equal(small))
+		Expect(s.summarizerModel()).To(Equal(small))
+		Expect(calls).To(Equal(1))
+	})
+
+	It("falls back to the pipeline model when the factory errors", func() {
+		pipeline := &fakeModel{}
+		s := &Session{ModelInterface: pipeline, SummaryModel: "tiny",
+			summarizerFactory: func() (Model, error) { return nil, errors.New("nope") }}
+		Expect(s.summarizerModel()).To(Equal(pipeline))
+	})
+})
+
+var _ = Describe("itemID", func() {
+	It("returns the id for each variant and empty for nil", func() {
+		Expect(itemID(nil)).To(Equal(""))
+		Expect(itemID(&types.MessageItemUnion{User: &types.MessageItemUser{ID: "u1"}})).To(Equal("u1"))
+		Expect(itemID(&types.MessageItemUnion{Assistant: &types.MessageItemAssistant{ID: "a1"}})).To(Equal("a1"))
+		Expect(itemID(&types.MessageItemUnion{System: &types.MessageItemSystem{ID: "s1"}})).To(Equal("s1"))
+		Expect(itemID(&types.MessageItemUnion{FunctionCall: &types.MessageItemFunctionCall{ID: "f1"}})).To(Equal("f1"))
+		Expect(itemID(&types.MessageItemUnion{FunctionCallOutput: &types.MessageItemFunctionCallOutput{ID: "o1"}})).To(Equal("o1"))
+	})
+})

core/http/react-ui/e2e/model-config.spec.js

@@ -288,6 +288,21 @@ test.describe('Model Editor - Interactive Tab', () => {
     await expect(page.locator('input[placeholder^="match,"]')).toBeVisible()
   })
 
+  test('pattern min_len clamps a directly-typed negative to 0', async ({ page }) => {
+    const searchInput = page.locator('input[placeholder="Search fields to add..."]')
+    await searchInput.fill('Custom Secret Patterns')
+    const dropdown = searchInput.locator('..').locator('..')
+    await dropdown.locator('div', { hasText: 'Custom Secret Patterns' }).first().click()
+
+    await page.locator('button', { hasText: 'Add pattern' }).click()
+    // The number input's min={0} only limits the spinner arrows, not keyboard
+    // entry; the editor must sanitise a typed negative so a meaningless
+    // negative length floor never reaches the saved config.
+    const minLen = page.locator('input[aria-label="Minimum length"]')
+    await minLen.fill('-5')
+    await expect(minLen).toHaveValue('0')
+  })
+
   // Regression: a map-typed field (entity_actions) present in the loaded YAML
   // must render WITH its values. flattenConfig used to recurse into the map,
   // scattering it across pii_detection.entity_actions.<GROUP> paths that match
@@ -329,4 +344,37 @@ test.describe('Model Editor - Interactive Tab', () => {
     await expect(page.getByText(/block —/i).first()).toBeVisible()
   })
 
+  // A map cannot hold two values for one key, so renaming a row to an existing
+  // group must collapse to a single row (Object.fromEntries, last write wins)
+  // rather than rendering two conflicting rows that silently lose one on save.
+  test('entity_actions collapses a duplicate group to a single row', async ({ page }) => {
+    await page.route('**/api/models/edit/ner-model', (route) => {
+      route.fulfill({
+        contentType: 'application/json',
+        body: JSON.stringify({
+          name: 'ner-model',
+          config: [
+            'name: ner-model',
+            'backend: llama-cpp',
+            'pii_detection:',
+            '    entity_actions:',
+            '        SSN: block',
+            '        EMAIL: mask',
+            '',
+          ].join('\n'),
+        }),
+      })
+    })
+
+    await page.goto('/app/model-editor/ner-model')
+
+    const groupInputs = page.locator('input[aria-label="Entity group"]')
+    await expect(groupInputs).toHaveCount(2)
+
+    // Rename the EMAIL row to duplicate SSN; the editor collapses to one SSN row.
+    await groupInputs.nth(1).fill('SSN')
+    await expect(groupInputs).toHaveCount(1)
+    await expect(groupInputs.nth(0)).toHaveValue('SSN')
+  })
+
 })

core/http/react-ui/e2e/nodes-detail.spec.js

@@ -0,0 +1,34 @@
+import { test, expect } from './coverage-fixtures.js'
+
+const ID = 'n1'
+async function mockNode(page) {
+  await page.route(`**/api/nodes/${ID}`, r => r.fulfill({ status: 200, contentType: 'application/json',
+    body: JSON.stringify({ id: ID, name: 'alpha', node_type: 'backend', address: '10.0.0.1:50051', status: 'healthy', total_vram: 24e9, available_vram: 12e9, max_replicas_per_model: 1, labels: { env: 'prod' } }) }))
+  await page.route(`**/api/nodes/${ID}/models`, r => r.fulfill({ status: 200, contentType: 'application/json',
+    body: JSON.stringify([{ node_id: ID, model_name: 'llama-3.3', state: 'loaded', in_flight: 0, replica_index: 0 }]) }))
+  await page.route(`**/api/nodes/${ID}/backends`, r => r.fulfill({ status: 200, contentType: 'application/json',
+    body: JSON.stringify([{ name: 'llama-cpp', is_system: true, installed_at: '2026-06-01T00:00:00Z' }]) }))
+}
+
+test.describe('Node detail page', () => {
+  test('renders sections for a node', async ({ page }) => {
+    await mockNode(page)
+    await page.goto(`/app/nodes/${ID}`)
+    await expect(page.locator('.page-title').first()).toBeVisible({ timeout: 15_000 })
+    await expect(page.getByText('alpha')).toBeVisible()
+    await expect(page.getByText('llama-3.3')).toBeVisible()
+    await expect(page.getByText('llama-cpp')).toBeVisible()
+    await expect(page.getByText('env=prod')).toBeVisible()
+  })
+
+  test('is reachable by clicking a roster panel', async ({ page }) => {
+    await page.route('**/api/nodes', r => r.fulfill({ status: 200, contentType: 'application/json',
+      body: JSON.stringify([{ id: ID, name: 'alpha', node_type: 'backend', address: '10.0.0.1:50051', status: 'healthy' }]) }))
+    await page.route('**/api/nodes/models', r => r.fulfill({ status: 200, contentType: 'application/json', body: '[]' }))
+    await page.route('**/api/nodes/scheduling', r => r.fulfill({ status: 200, contentType: 'application/json', body: '[]' }))
+    await mockNode(page)
+    await page.goto('/app/nodes')
+    await page.locator('.node-panel').filter({ hasText: 'alpha' }).getByText('alpha').click()
+    await expect(page).toHaveURL(new RegExp(`/app/nodes/${ID}$`))
+  })
+})

core/http/react-ui/e2e/nodes-per-node-backend-actions.spec.js

@@ -12,28 +12,37 @@ const NODE_NAME = 'worker-test'
 const BACKEND_NAME = 'cuda12-vllm-development'
 
 async function mockDistributedNodes(page, { onDelete } = {}) {
+  const nodeRecord = {
+    id: NODE_ID,
+    name: NODE_NAME,
+    node_type: 'backend',
+    address: '10.0.0.1:50051',
+    http_address: '10.0.0.1:8090',
+    status: 'healthy',
+    total_vram: 0,
+    available_vram: 0,
+    total_ram: 8_000_000_000,
+    available_ram: 4_000_000_000,
+    gpu_vendor: '',
+    last_heartbeat: new Date().toISOString(),
+    created_at: new Date().toISOString(),
+    updated_at: new Date().toISOString(),
+  }
+
   await page.route('**/api/nodes', (route) => {
     route.fulfill({
       status: 200,
       contentType: 'application/json',
-      body: JSON.stringify([
-        {
-          id: NODE_ID,
-          name: NODE_NAME,
-          node_type: 'backend',
-          address: '10.0.0.1:50051',
-          http_address: '10.0.0.1:8090',
-          status: 'healthy',
-          total_vram: 0,
-          available_vram: 0,
-          total_ram: 8_000_000_000,
-          available_ram: 4_000_000_000,
-          gpu_vendor: '',
-          last_heartbeat: new Date().toISOString(),
-          created_at: new Date().toISOString(),
-          updated_at: new Date().toISOString(),
-        },
-      ]),
+      body: JSON.stringify([nodeRecord]),
+    })
+  })
+
+  // The detail page fetches the single node via nodesApi.get(id).
+  await page.route(`**/api/nodes/${NODE_ID}`, (route) => {
+    route.fulfill({
+      status: 200,
+      contentType: 'application/json',
+      body: JSON.stringify(nodeRecord),
     })
   })
 
@@ -80,24 +89,18 @@ async function mockDistributedNodes(page, { onDelete } = {}) {
   })
 }
 
-async function expandNodeAndWaitForBackends(page) {
-  await page.goto('/app/nodes')
-  // Click the row to expand it. The chevron toggle and the row both work,
-  // but clicking the name cell is the most user-like.
-  await page.getByText(NODE_NAME).first().click()
-  // Backends, Capacity and Labels live behind a "Manage" <details>
-  // disclosure (the drawer was distilled to keep at-a-glance content
-  // lean — see distill refactor in the multi-replica branch). Open it
-  // by clicking the summary inside the .node-manage scope so the
-  // per-node backend table is in the DOM before assertions run.
-  await page.locator('.node-manage > summary').first().click()
+async function openNodeDetail(page) {
+  // The per-node backend table now lives on the deep-linkable detail page
+  // at /app/nodes/:id (the old expand-row + "Manage" disclosure was removed
+  // when the roster was restructured). Navigate straight there.
+  await page.goto(`/app/nodes/${NODE_ID}`)
   await expect(page.getByRole('cell', { name: BACKEND_NAME, exact: true })).toBeVisible({ timeout: 10_000 })
 }
 
 test.describe('Nodes page — per-node backend actions', () => {
   test('upgrade affordance is self-explanatory (not "Reinstall backend" with a sync icon)', async ({ page }) => {
     await mockDistributedNodes(page)
-    await expandNodeAndWaitForBackends(page)
+    await openNodeDetail(page)
 
     // Negative: the old, ambiguous wording must not be used.
     await expect(page.locator('button[title="Reinstall backend"]')).toHaveCount(0)
@@ -114,7 +117,7 @@ test.describe('Nodes page — per-node backend actions', () => {
 
   test('per-node backend row shows a delete (trash) button next to upgrade', async ({ page }) => {
     await mockDistributedNodes(page)
-    await expandNodeAndWaitForBackends(page)
+    await openNodeDetail(page)
 
     const deleteBtn = page.locator('button[title="Delete backend from this node"]')
     await expect(deleteBtn).toBeVisible()
@@ -128,7 +131,7 @@ test.describe('Nodes page — per-node backend actions', () => {
         postedBody = route.request().postDataJSON()
       },
     })
-    await expandNodeAndWaitForBackends(page)
+    await openNodeDetail(page)
 
     await page.locator('button[title="Delete backend from this node"]').click()
 
@@ -150,7 +153,7 @@ test.describe('Nodes page — per-node backend actions', () => {
         deleteCalls += 1
       },
     })
-    await expandNodeAndWaitForBackends(page)
+    await openNodeDetail(page)
 
     await page.locator('button[title="Delete backend from this node"]').click()
 

core/http/react-ui/e2e/nodes-roster.spec.js

@@ -0,0 +1,47 @@
+import { test, expect } from './coverage-fixtures.js'
+
+async function mockCluster(page, nodes) {
+  await page.route('**/api/nodes', r => r.fulfill({ status: 200, contentType: 'application/json', body: JSON.stringify(nodes) }))
+  await page.route('**/api/nodes/models', r => r.fulfill({ status: 200, contentType: 'application/json', body: '[]' }))
+  await page.route('**/api/nodes/scheduling', r => r.fulfill({ status: 200, contentType: 'application/json', body: '[]' }))
+}
+
+test.describe('Nodes roster header', () => {
+  test('shows a cluster pulse line and no stat-card grid', async ({ page }) => {
+    await mockCluster(page, [
+      { id: 'n1', name: 'alpha', node_type: 'backend', address: '10.0.0.1:50051', status: 'healthy' },
+      { id: 'n2', name: 'beta', node_type: 'backend', address: '10.0.0.2:50051', status: 'draining' },
+    ])
+    await page.goto('/app/nodes')
+    await expect(page.locator('.cluster-pulse')).toBeVisible({ timeout: 15_000 })
+    await expect(page.locator('.cluster-pulse')).toContainText('2 nodes')
+    await expect(page.locator('.stat-grid')).toHaveCount(0)
+  })
+
+  test('shows an approval callout for pending nodes', async ({ page }) => {
+    await mockCluster(page, [{ id: 'n3', name: 'gamma', node_type: 'backend', address: '10.0.0.3:50051', status: 'pending' }])
+    await page.goto('/app/nodes')
+    await expect(page.locator('.attention-callout')).toContainText('approval', { timeout: 15_000 })
+  })
+})
+
+test.describe('Nodes roster panels', () => {
+  test('shows model chips without clicking and filters by type', async ({ page }) => {
+    await page.route('**/api/nodes', r => r.fulfill({ status: 200, contentType: 'application/json', body: JSON.stringify([
+      { id: 'n1', name: 'alpha', node_type: 'backend', address: '10.0.0.1:50051', status: 'healthy' },
+      { id: 'a1', name: 'agent-1', node_type: 'agent', address: '10.0.0.9:50051', status: 'healthy' },
+    ]) }))
+    await page.route('**/api/nodes/models', r => r.fulfill({ status: 200, contentType: 'application/json', body: JSON.stringify([
+      { node_id: 'n1', model_name: 'llama-3.3', state: 'loaded', in_flight: 2, replica_index: 0 },
+    ]) }))
+    await page.route('**/api/nodes/scheduling', r => r.fulfill({ status: 200, contentType: 'application/json', body: '[]' }))
+
+    await page.goto('/app/nodes')
+    // model chip visible without any expand click
+    await expect(page.locator('.node-panel').filter({ hasText: 'alpha' }).getByText('llama-3.3')).toBeVisible({ timeout: 15_000 })
+    // segmented filter: Agent shows the agent node, hides the backend node
+    await page.getByRole('radio', { name: /Agent/ }).click()
+    await expect(page.getByText('agent-1')).toBeVisible()
+    await expect(page.getByText('alpha')).toHaveCount(0)
+  })
+})

core/http/react-ui/e2e/page-render-smoke.spec.js

@@ -21,6 +21,7 @@ const PAGES = [
   ['/app/backends', 'Backends'],
   ['/app/settings', 'Settings'],
   ['/app/nodes', 'Nodes'],
+  ['/app/scheduling', 'Scheduling'],
   ['/app/face', 'Face recognition'],
   ['/app/voice', 'Voice recognition'],
   ['/app/fine-tune', 'Fine-tuning'],

core/http/react-ui/e2e/scheduling.spec.js

@@ -0,0 +1,16 @@
+import { test, expect } from './coverage-fixtures.js'
+
+test.describe('Scheduling page', () => {
+  test('renders at /app/scheduling with rules from the API', async ({ page }) => {
+    await page.route('**/api/nodes/scheduling', (route) => {
+      route.fulfill({
+        status: 200, contentType: 'application/json',
+        body: JSON.stringify([{ model_name: 'llama-3.3', spread_all: true, min_replicas: 0, max_replicas: 0 }]),
+      })
+    })
+    await page.goto('/app/scheduling')
+    await expect(page.locator('.page-title').first()).toBeVisible({ timeout: 15_000 })
+    await expect(page).toHaveURL(/\/app\/scheduling$/)
+    await expect(page.getByText('llama-3.3')).toBeVisible()
+  })
+})

core/http/react-ui/public/locales/de/admin.json

@@ -43,6 +43,10 @@
     "title": "Verteilte Knoten",
     "subtitle": "Backend- und Agenten-Worker-Knoten verwalten"
   },
+  "scheduling": {
+    "title": "Planung",
+    "subtitle": "Modellplatzierung und Replikat-Regeln im gesamten Cluster"
+  },
   "p2p": {
     "title": "Verteilte KI-Berechnung",
     "subtitle": "Skalieren Sie Ihre KI-Workloads über mehrere Geräte mit Peer-to-Peer-Verteilung"

core/http/react-ui/public/locales/de/nav.json

@@ -50,6 +50,7 @@
     "backends": "Backends",
     "traces": "Traces",
     "nodes": "Knoten",
+    "scheduling": "Planung",
     "swarm": "Swarm",
     "system": "System",
     "settings": "Einstellungen",

core/http/react-ui/public/locales/en/admin.json

@@ -43,6 +43,10 @@
     "title": "Distributed Nodes",
     "subtitle": "Manage backend and agent worker nodes"
   },
+  "scheduling": {
+    "title": "Scheduling",
+    "subtitle": "Model placement and replica rules across the cluster"
+  },
   "p2p": {
     "title": "Distributed AI Computing",
     "subtitle": "Scale your AI workloads across multiple devices with peer-to-peer distribution"

core/http/react-ui/public/locales/en/nav.json

@@ -51,6 +51,7 @@
     "backends": "Backends",
     "traces": "Traces",
     "nodes": "Nodes",
+    "scheduling": "Scheduling",
     "swarm": "Swarm",
     "system": "System",
     "settings": "Settings",

core/http/react-ui/public/locales/es/admin.json

@@ -43,6 +43,10 @@
     "title": "Nodos distribuidos",
     "subtitle": "Administra nodos worker de backends y agentes"
   },
+  "scheduling": {
+    "title": "Planificación",
+    "subtitle": "Reglas de ubicación de modelos y réplicas en el clúster"
+  },
   "p2p": {
     "title": "Computación de IA distribuida",
     "subtitle": "Escala tus cargas de trabajo de IA en múltiples dispositivos con distribución peer-to-peer"

core/http/react-ui/public/locales/es/nav.json

@@ -50,6 +50,7 @@
     "backends": "Backends",
     "traces": "Trazas",
     "nodes": "Nodos",
+    "scheduling": "Planificación",
     "swarm": "Swarm",
     "system": "Sistema",
     "settings": "Configuración",

core/http/react-ui/public/locales/id/admin.json

@@ -43,6 +43,10 @@
     "title": "Node Terdistribusi",
     "subtitle": "Kelola node backend dan node worker"
   },
+  "scheduling": {
+    "title": "Penjadwalan",
+    "subtitle": "Aturan penempatan model dan replika di seluruh klaster"
+  },
   "p2p": {
     "title": "Komputasi AI Terdistribusi",
     "subtitle": "Skalakan beban kerja AI Anda ke beberapa perangkat dengan distribusi peer-to-peer"

core/http/react-ui/public/locales/id/nav.json

@@ -51,6 +51,7 @@
     "backends": "Backend",
     "traces": "Trace",
     "nodes": "Node",
+    "scheduling": "Penjadwalan",
     "swarm": "Swarm",
     "system": "Sistem",
     "settings": "Pengaturan",

core/http/react-ui/public/locales/it/admin.json

@@ -43,6 +43,10 @@
     "title": "Nodi distribuiti",
     "subtitle": "Gestisci i nodi worker dei backend e degli agenti"
   },
+  "scheduling": {
+    "title": "Pianificazione",
+    "subtitle": "Regole di posizionamento dei modelli e delle repliche nel cluster"
+  },
   "p2p": {
     "title": "Calcolo AI distribuito",
     "subtitle": "Scala i tuoi carichi di lavoro AI su più dispositivi con la distribuzione peer-to-peer"

core/http/react-ui/public/locales/it/nav.json

@@ -50,6 +50,7 @@
     "backends": "Backend",
     "traces": "Tracce",
     "nodes": "Nodi",
+    "scheduling": "Pianificazione",
     "swarm": "Swarm",
     "system": "Sistema",
     "settings": "Impostazioni",

core/http/react-ui/public/locales/ko/admin.json

@@ -43,6 +43,10 @@
     "title": "분산 노드",
     "subtitle": "백엔드 및 에이전트 워커 노드를 관리합니다"
   },
+  "scheduling": {
+    "title": "스케줄링",
+    "subtitle": "클러스터 전반의 모델 배치 및 복제본 규칙"
+  },
   "p2p": {
     "title": "분산 AI 컴퓨팅",
     "subtitle": "피어 투 피어 분산으로 여러 기기에 걸쳐 AI 워크로드를 확장합니다"

core/http/react-ui/public/locales/ko/nav.json

@@ -51,6 +51,7 @@
     "backends": "백엔드",
     "traces": "트레이스",
     "nodes": "노드",
+    "scheduling": "스케줄링",
     "swarm": "Swarm",
     "system": "시스템",
     "settings": "설정",

core/http/react-ui/public/locales/zh-CN/admin.json

@@ -43,6 +43,10 @@
     "title": "分布式节点",
     "subtitle": "管理后端和智能体工作节点"
   },
+  "scheduling": {
+    "title": "调度",
+    "subtitle": "集群中的模型放置和副本规则"
+  },
   "p2p": {
     "title": "分布式 AI 计算",
     "subtitle": "通过点对点分发将您的 AI 工作负载扩展到多个设备"

core/http/react-ui/public/locales/zh-CN/nav.json

@@ -50,6 +50,7 @@
     "backends": "后端",
     "traces": "追踪",
     "nodes": "节点",
+    "scheduling": "调度",
     "swarm": "Swarm",
     "system": "系统",
     "settings": "设置",

core/http/react-ui/src/App.css

@@ -8471,3 +8471,56 @@ select.input {
 .status-pill--error   .status-pill__dot { background: var(--color-error); }
 .status-pill--info    .status-pill__dot { background: var(--color-info); }
 .status-pill--muted   .status-pill__dot { background: var(--color-text-muted); }
+
+/* Nodes: cluster pulse + attention callout (replaces the stat-card strip) */
+.cluster-pulse {
+  font-size: var(--text-sm);
+  color: var(--color-text-muted);
+  margin: 0 0 var(--spacing-lg);
+}
+.cluster-pulse__strong { color: var(--color-text-primary); font-weight: 600; }
+
+.attention-callout {
+  display: flex;
+  align-items: center;
+  justify-content: space-between;
+  gap: var(--spacing-md);
+  padding: var(--spacing-sm) var(--spacing-md);
+  border-radius: var(--radius-md);
+  margin-bottom: var(--spacing-lg);
+  font-size: var(--text-sm);
+}
+.attention-callout--warn {
+  background: var(--color-warning-light);
+  border: 1px solid var(--color-warning-border);
+  color: var(--color-text-primary);
+}
+.attention-callout--error {
+  background: var(--color-error-light);
+  border: 1px solid var(--color-error-border);
+  color: var(--color-text-primary);
+}
+
+/* Node roster panels (Nodes page) */
+.node-roster { display: flex; flex-direction: column; gap: var(--spacing-sm); }
+.node-panel {
+  background: var(--color-bg-secondary);
+  border: 1px solid var(--color-border-subtle);
+  border-radius: var(--radius-lg);
+}
+.node-panel__main { padding: var(--spacing-md) var(--spacing-lg); cursor: pointer; }
+.node-panel:hover { border-color: var(--color-border); }
+.node-panel__head { display: flex; align-items: flex-start; justify-content: space-between; gap: var(--spacing-md); }
+.node-panel__id { display: flex; align-items: center; gap: var(--spacing-sm); flex-wrap: wrap; }
+.node-panel__name { font-weight: 600; }
+.node-panel__meta { display: flex; gap: var(--spacing-lg); margin-top: var(--spacing-sm); color: var(--color-text-muted); font-size: var(--text-xs); }
+.node-panel__models { display: flex; flex-wrap: wrap; gap: 6px; margin-top: var(--spacing-sm); }
+.model-chip {
+  display: inline-flex; align-items: center; gap: 5px;
+  font-family: var(--font-mono); font-size: 0.6875rem;
+  padding: 2px 8px; border-radius: var(--radius-sm); border: 1px solid;
+}
+.model-chip__dot { width: 6px; height: 6px; border-radius: 50%; }
+.model-chip__state { opacity: 0.85; font-style: normal; }
+.node-filter { margin-bottom: var(--spacing-lg); }
+.node-detail__metrics { display: flex; gap: var(--spacing-xl); margin: var(--spacing-md) 0 var(--spacing-lg); flex-wrap: wrap; }

core/http/react-ui/src/components/PatternListEditor.jsx

@@ -74,7 +74,18 @@ export default function PatternListEditor({ value, onChange }) {
             min={0}
             value={r.min_len || 0}
             title="Minimum match length (0 = no floor)"
-            onChange={e => update(i, { min_len: parseInt(e.target.value, 10) || 0 })}
+            // min={0} only constrains the spinner, not keyboard entry. Clamp a
+            // typed negative to 0 (a negative floor is meaningless and would
+            // disable the length filter). When we clamp, force the DOM value
+            // too: the resulting 0->0 state change is a no-op, so React's
+            // controlled input would otherwise keep displaying the rejected
+            // "-5" even though the saved value is 0.
+            onChange={e => {
+              const parsed = parseInt(e.target.value, 10)
+              const n = Math.max(0, parsed || 0)
+              if (parsed < 0) e.target.value = String(n)
+              update(i, { min_len: n })
+            }}
             style={{ width: 80, fontSize: '0.8125rem' }}
             aria-label="Minimum length"
           />

core/http/react-ui/src/components/console/consoleConfig.js

@@ -59,6 +59,7 @@ export const operateConsole = {
       titleKey: 'operate.cluster',
       items: [
         { path: '/app/nodes', icon: 'fas fa-network-wired', labelKey: 'items.nodes', adminOnly: true, feature: 'distributed' },
+        { path: '/app/scheduling', icon: 'fas fa-calendar-alt', labelKey: 'items.scheduling', adminOnly: true, feature: 'distributed' },
         { path: '/app/p2p', icon: 'fas fa-circle-nodes', labelKey: 'items.swarm', adminOnly: true },
       ],
     },

core/http/react-ui/src/components/nodes/AttentionCallout.jsx

@@ -0,0 +1,31 @@
+export default function AttentionCallout({ nodes, onApprove }) {
+  const pending = nodes.filter(n => n.status === 'pending')
+  const unhealthy = nodes.filter(n => n.status === 'unhealthy' || n.status === 'offline')
+  if (pending.length === 0 && unhealthy.length === 0) return null
+
+  if (pending.length > 0) {
+    const first = pending[0]
+    const extra = pending.length - 1
+    return (
+      <div className="attention-callout attention-callout--warn">
+        <span>
+          <i className="fas fa-exclamation-circle" />{' '}
+          <strong>{pending.length} node{pending.length > 1 ? 's' : ''} awaiting approval</strong>
+          {' - '}{first.name}{extra > 0 ? ` +${extra} more` : ''}
+        </span>
+        <button className="btn btn-primary btn-sm" onClick={() => onApprove(first.id)}>
+          <i className="fas fa-check" /> Approve {first.name}
+        </button>
+      </div>
+    )
+  }
+  return (
+    <div className="attention-callout attention-callout--error">
+      <span>
+        <i className="fas fa-exclamation-triangle" />{' '}
+        <strong>{unhealthy.length} node{unhealthy.length > 1 ? 's' : ''} unhealthy</strong>
+        {' - '}{unhealthy.map(n => n.name).slice(0, 3).join(', ')}
+      </span>
+    </div>
+  )
+}

core/http/react-ui/src/components/nodes/CapacityEditor.jsx

@@ -0,0 +1,196 @@
+import { useState, useEffect, useCallback } from 'react'
+import { nodesApi } from '../../utils/api'
+import LoadingSpinner from '../LoadingSpinner'
+
+/**
+ * Inline editor for a node's per-model replica capacity.
+ *
+ * UX intent: discoverable affordance (pencil icon) that opens an inline
+ * input - never a modal for a single field. Source-of-truth note is shown
+ * inline so operators understand a worker re-registration will overwrite
+ * their override; surfacing this in a tooltip would hide too important a
+ * caveat.
+ *
+ * `confirmShrink` is a hook the parent provides so the page can render its
+ * own confirm dialog (it has access to all nodes and can phrase the message
+ * with full context).
+ */
+export default function CapacityEditor({ node, loadedModelCounts, onUpdate, confirmShrink, addToast }) {
+  const current = node.max_replicas_per_model || 1
+  const isOverride = !!node.max_replicas_per_model_manually_set
+  const [editing, setEditing] = useState(false)
+  const [draft, setDraft] = useState(String(current))
+  const [saving, setSaving] = useState(false)
+  const [resetting, setResetting] = useState(false)
+
+  // Reset draft when current value changes (server response, etc.)
+  useEffect(() => {
+    if (!editing) setDraft(String(current))
+  }, [current, editing])
+
+  const cancel = useCallback(() => {
+    setEditing(false)
+    setDraft(String(current))
+  }, [current])
+
+  const save = useCallback(async () => {
+    const value = parseInt(draft, 10)
+    if (!Number.isFinite(value) || value < 1) {
+      addToast('Replica capacity must be 1 or higher', 'error')
+      return
+    }
+    if (value === current) {
+      setEditing(false)
+      return
+    }
+    // Reducing the cap below current loaded replicas: confirm so the operator
+    // sees the consequence (running replicas keep going until idle eviction).
+    const maxLoadedAcrossModels = Math.max(0, ...Object.values(loadedModelCounts || {}))
+    if (value < maxLoadedAcrossModels) {
+      const proceed = await confirmShrink({ node, newValue: value, currentLoaded: maxLoadedAcrossModels })
+      if (!proceed) return
+    }
+    setSaving(true)
+    try {
+      await nodesApi.updateMaxReplicasPerModel(node.id, value)
+      addToast(`Replica capacity set to ${value} on ${node.name}`, 'success')
+      setEditing(false)
+      onUpdate?.(value)
+    } catch (err) {
+      addToast(`Could not change replica capacity: ${err.message || err}`, 'error')
+    } finally {
+      setSaving(false)
+    }
+  }, [draft, current, node, loadedModelCounts, confirmShrink, onUpdate, addToast])
+
+  const onKeyDown = (e) => {
+    if (e.key === 'Enter') { e.preventDefault(); save() }
+    else if (e.key === 'Escape') { e.preventDefault(); cancel() }
+  }
+
+  const reset = useCallback(async () => {
+    setResetting(true)
+    try {
+      await nodesApi.resetMaxReplicasPerModel(node.id)
+      addToast(`Override cleared on ${node.name}; worker flag will apply on next re-registration`, 'success')
+      onUpdate?.(null)
+    } catch (err) {
+      addToast(`Could not reset override: ${err.message || err}`, 'error')
+    } finally {
+      setResetting(false)
+    }
+  }, [node, onUpdate, addToast])
+
+  return (
+    <div style={{
+      display: 'flex', alignItems: 'flex-start', gap: 'var(--spacing-md)',
+    }}>
+      <i className="fas fa-layer-group" style={{ color: 'var(--color-text-muted)', marginTop: 3 }} aria-hidden="true" />
+      <div style={{ flex: 1, minWidth: 0 }}>
+        <div style={{ display: 'flex', alignItems: 'center', gap: 'var(--spacing-sm)', flexWrap: 'wrap' }}>
+          <label
+            htmlFor={`capacity-${node.id}`}
+            style={{ fontSize: '0.8125rem', fontWeight: 600, color: 'var(--color-text-primary)' }}
+          >
+            Max replicas per model
+          </label>
+          {editing ? (
+            <>
+              <input
+                id={`capacity-${node.id}`}
+                type="number"
+                min={1}
+                value={draft}
+                disabled={saving}
+                onChange={(e) => setDraft(e.target.value)}
+                onKeyDown={onKeyDown}
+                autoFocus
+                aria-describedby={`capacity-hint-${node.id}`}
+                style={{
+                  width: 72, padding: '4px 8px', borderRadius: 'var(--radius-sm)',
+                  border: '1px solid var(--color-border)', background: 'var(--color-bg-primary)',
+                  fontFamily: 'var(--font-mono)', fontSize: '0.8125rem',
+                  color: 'var(--color-text-primary)',
+                }}
+              />
+              <button
+                className="btn btn-primary btn-sm"
+                onClick={save}
+                disabled={saving}
+                style={{ minHeight: 32 }}
+                aria-label="Save replica capacity"
+              >
+                {saving ? <LoadingSpinner size="xs" /> : <><i className="fas fa-check" /> Save</>}
+              </button>
+              <button
+                className="btn btn-secondary btn-sm"
+                onClick={cancel}
+                disabled={saving}
+                style={{ minHeight: 32 }}
+                aria-label="Cancel"
+              >
+                Cancel
+              </button>
+            </>
+          ) : (
+            <>
+              <span
+                className="cell-mono"
+                style={{ fontSize: '0.8125rem', color: 'var(--color-text-secondary)' }}
+              >
+                {current}
+              </span>
+              {isOverride && (
+                <span
+                  title="This value was set from the UI. It will persist across worker restarts until you click Reset."
+                  style={{
+                    display: 'inline-block', fontSize: '0.6875rem', padding: '1px 6px',
+                    borderRadius: 'var(--radius-sm)', fontWeight: 500,
+                    background: 'var(--color-bg-primary)',
+                    border: '1px solid var(--color-warning, #d97706)',
+                    color: 'var(--color-warning, #d97706)',
+                  }}
+                >
+                  override
+                </span>
+              )}
+              <button
+                onClick={() => setEditing(true)}
+                aria-label={`Edit replica capacity (currently ${current})`}
+                title="Change replica capacity for this node"
+                style={{
+                  display: 'inline-flex', alignItems: 'center', justifyContent: 'center',
+                  minWidth: 32, minHeight: 32, padding: 4, borderRadius: 'var(--radius-sm)',
+                  border: '1px solid var(--color-border-subtle)',
+                  background: 'transparent', color: 'var(--color-text-muted)', cursor: 'pointer',
+                }}
+              >
+                <i className="fas fa-pencil-alt" />
+              </button>
+              {isOverride && (
+                <button
+                  onClick={reset}
+                  disabled={resetting}
+                  aria-label="Clear admin override and let the worker flag apply"
+                  title="Clear override; the worker's --max-replicas-per-model flag will apply on the next re-registration"
+                  className="btn btn-secondary btn-sm"
+                  style={{ minHeight: 32 }}
+                >
+                  {resetting ? <LoadingSpinner size="xs" /> : <><i className="fas fa-undo" /> Reset</>}
+                </button>
+              )}
+            </>
+          )}
+        </div>
+        <div
+          id={`capacity-hint-${node.id}`}
+          style={{ fontSize: '0.75rem', color: 'var(--color-text-muted)', marginTop: 4, lineHeight: 1.4 }}
+        >
+          {isOverride
+            ? <>Set from here. <strong>Reset</strong> to use the worker's default.</>
+            : <>Saved values stick across worker restarts.</>}
+        </div>
+      </div>
+    </div>
+  )
+}

core/http/react-ui/src/components/nodes/ClusterPulse.jsx

@@ -0,0 +1,18 @@
+import { formatVRAM } from './nodeStatus'
+
+export default function ClusterPulse({ nodes }) {
+  const total = nodes.length
+  const healthy = nodes.filter(n => n.status === 'healthy').length
+  const draining = nodes.filter(n => n.status === 'draining').length
+  const usedVRAM = nodes.reduce((s, n) =>
+    (n.total_vram && n.available_vram != null) ? s + (n.total_vram - n.available_vram) : s, 0)
+  const vramStr = formatVRAM(usedVRAM)
+  return (
+    <p className="cluster-pulse">
+      <span className="cluster-pulse__strong">{total} {total === 1 ? 'node' : 'nodes'}</span>
+      {' · '}<span style={{ color: 'var(--color-success)' }}>{healthy} healthy</span>
+      {draining > 0 && <>{' · '}<span style={{ color: 'var(--color-warning)' }}>{draining} draining</span></>}
+      {vramStr && <>{' · '}{vramStr} VRAM in use</>}
+    </p>
+  )
+}

core/http/react-ui/src/components/nodes/KeyValueChips.jsx

@@ -0,0 +1,98 @@
+import { useState } from 'react'
+
+/**
+ * Controlled chip-builder for { key: value } maps. Replaces the prior
+ * comma-separated-string Node Selector input AND the bespoke Labels editor
+ * in the node drawer - both were rendering the same chip pattern with
+ * subtly different markup.
+ *
+ * Fully controlled: parent owns the map and decides what onAdd/onRemove
+ * does (form state for the scheduling form; API calls for the live
+ * labels editor). The component just renders chips and a key/value input
+ * row.
+ *
+ * Props:
+ *   pairs       - current map of key -> value
+ *   onAdd(k,v)  - called when the user adds a pair (parent handles dedup
+ *                 and persistence side effects)
+ *   onRemove(k) - called when a chip's × is clicked
+ *   placeholderKey, placeholderValue - input hints
+ *   ariaLabel   - accessible name for the section
+ */
+export default function KeyValueChips({ pairs, onAdd, onRemove, placeholderKey = 'key', placeholderValue = 'value', ariaLabel }) {
+  const [k, setK] = useState('')
+  const [v, setV] = useState('')
+
+  const add = () => {
+    const key = k.trim()
+    if (!key) return
+    onAdd(key, v.trim())
+    setK(''); setV('')
+  }
+  const onKeyDown = (e) => {
+    if (e.key === 'Enter') { e.preventDefault(); add() }
+  }
+
+  const entries = pairs ? Object.entries(pairs) : []
+  return (
+    <div aria-label={ariaLabel}>
+      {entries.length > 0 && (
+        <div style={{ display: 'flex', flexWrap: 'wrap', gap: 4, marginBottom: 'var(--spacing-xs)' }}>
+          {entries.map(([key, val]) => (
+            <span key={key} style={{
+              display: 'inline-flex', alignItems: 'center', gap: 4,
+              fontSize: '0.75rem', padding: '2px 8px',
+              borderRadius: 'var(--radius-sm)',
+              background: 'var(--color-bg-tertiary)',
+              border: '1px solid var(--color-border-subtle)',
+              fontFamily: 'var(--font-mono)',
+            }}>
+              {key}={val}
+              <button
+                type="button"
+                onClick={(e) => { e.stopPropagation(); onRemove(key) }}
+                aria-label={`Remove ${key}`}
+                title="Remove"
+                style={{
+                  background: 'none', border: 'none', cursor: 'pointer',
+                  color: 'var(--color-text-muted)', fontSize: '0.625rem', padding: 0,
+                }}
+              >
+                <i className="fas fa-times" />
+              </button>
+            </span>
+          ))}
+        </div>
+      )}
+      <div style={{ display: 'flex', gap: 'var(--spacing-xs)', alignItems: 'stretch' }}>
+        <input
+          className="input"
+          type="text"
+          placeholder={placeholderKey}
+          value={k}
+          onChange={e => setK(e.target.value)}
+          onKeyDown={onKeyDown}
+          style={{ flex: 1 }}
+        />
+        <input
+          className="input"
+          type="text"
+          placeholder={placeholderValue}
+          value={v}
+          onChange={e => setV(e.target.value)}
+          onKeyDown={onKeyDown}
+          style={{ flex: 1 }}
+        />
+        <button
+          type="button"
+          className="btn btn-secondary btn-sm"
+          onClick={add}
+          disabled={!k.trim()}
+          style={{ minHeight: 36 }}
+        >
+          <i className="fas fa-plus" /> Add
+        </button>
+      </div>
+    </div>
+  )
+}

core/http/react-ui/src/components/nodes/ModelChip.jsx

@@ -0,0 +1,12 @@
+import { modelStateConfig } from './nodeStatus'
+
+export default function ModelChip({ model }) {
+  const cfg = modelStateConfig[model.state] || modelStateConfig.idle
+  return (
+    <span className="model-chip" style={{ background: cfg.bg, color: cfg.color, borderColor: cfg.border }}>
+      <span className="model-chip__dot" style={{ background: cfg.color }} />
+      {model.model_name}
+      {model.state !== 'loaded' && <span className="model-chip__state"> {model.state}</span>}
+    </span>
+  )
+}

core/http/react-ui/src/components/nodes/NodePanel.jsx

@@ -0,0 +1,60 @@
+import { useNavigate } from 'react-router-dom'
+import StatusPill from './StatusPill'
+import ModelChip from './ModelChip'
+import ActionMenu from '../ActionMenu'
+import { formatVRAM } from './nodeStatus'
+
+export default function NodePanel({ node, models = [], onApprove, onDrain, onResume, onRemove }) {
+  const navigate = useNavigate()
+  const isAgent = node.node_type === 'agent'
+  const open = () => navigate(`/app/nodes/${node.id}`)
+  const usedVRAM = node.total_vram && node.available_vram != null ? node.total_vram - node.available_vram : null
+
+  return (
+    <div className="node-panel">
+      <div className="node-panel__main" onClick={open} role="button" tabIndex={0}
+        onKeyDown={(e) => { if (e.key === 'Enter') open() }}>
+        <div className="node-panel__head">
+          <div className="node-panel__id">
+            <StatusPill status={node.status} />
+            <span className="node-panel__name">{node.name}</span>
+            <span className="cell-mono cell-muted">{node.address}</span>
+          </div>
+          <div className="node-panel__actions" onClick={(e) => e.stopPropagation()}>
+            {node.status === 'pending' && (
+              <button className="btn btn-primary btn-sm" onClick={() => onApprove(node.id)}>
+                <i className="fas fa-check" /> Approve
+              </button>
+            )}
+            <ActionMenu
+              ariaLabel={`Actions for ${node.name}`}
+              triggerLabel={`Actions for ${node.name}`}
+              items={[
+                { key: 'resume', icon: 'fa-play', label: 'Resume', hidden: node.status !== 'draining', onClick: () => onResume(node.id) },
+                { key: 'drain', icon: 'fa-pause', label: 'Drain', hidden: node.status === 'draining' || node.status === 'pending', onClick: () => onDrain(node.id) },
+                { divider: true, hidden: node.status === 'pending' },
+                { key: 'remove', icon: 'fa-trash', label: 'Remove from cluster', danger: true, onClick: () => onRemove(node) },
+              ]}
+            />
+          </div>
+        </div>
+
+        {!isAgent && (
+          <>
+            <div className="node-panel__meta">
+              {node.total_vram > 0 && (
+                <span className="cell-mono">VRAM {formatVRAM(usedVRAM) || '0'} / {formatVRAM(node.total_vram)}</span>
+              )}
+              <span className="cell-mono">{node.in_flight_count || 0} in-flight</span>
+            </div>
+            <div className="node-panel__models">
+              {models.length === 0
+                ? <span className="cell-muted">No models loaded</span>
+                : models.map(m => <ModelChip key={`${m.model_name}-${m.replica_index ?? 0}`} model={m} />)}
+            </div>
+          </>
+        )}
+      </div>
+    </div>
+  )
+}

core/http/react-ui/src/components/nodes/StatusPill.jsx

@@ -0,0 +1,11 @@
+import { statusConfig } from './nodeStatus'
+
+export default function StatusPill({ status }) {
+  const cfg = statusConfig[status] || statusConfig.unhealthy
+  return (
+    <span className="node-status" style={{ color: cfg.color }}>
+      <span className="node-status__dot" style={{ background: cfg.color }} />
+      {cfg.label}
+    </span>
+  )
+}

core/http/react-ui/src/components/nodes/nodeStatus.js

@@ -0,0 +1,34 @@
+export const statusConfig = {
+  healthy: { color: 'var(--color-success)', label: 'Healthy' },
+  unhealthy: { color: 'var(--color-error)', label: 'Unhealthy' },
+  offline: { color: 'var(--color-error)', label: 'Offline' },
+  registering: { color: 'var(--color-primary)', label: 'Registering' },
+  draining: { color: 'var(--color-warning)', label: 'Draining' },
+  pending: { color: 'var(--color-warning)', label: 'Pending Approval' },
+}
+
+export const modelStateConfig = {
+  loaded: { bg: 'var(--color-success-light)', color: 'var(--color-success)', border: 'var(--color-success-border)' },
+  loading: { bg: 'var(--color-primary-light)', color: 'var(--color-primary)', border: 'var(--color-primary-border)' },
+  unloading: { bg: 'var(--color-warning-light)', color: 'var(--color-warning)', border: 'var(--color-warning-border)' },
+  idle: { bg: 'var(--color-bg-tertiary)', color: 'var(--color-text-muted)', border: 'var(--color-border-subtle)' },
+}
+
+export function formatVRAM(bytes) {
+  if (!bytes || bytes === 0) return null
+  const gb = bytes / (1024 * 1024 * 1024)
+  return gb >= 1 ? `${gb.toFixed(1)} GB` : `${(bytes / (1024 * 1024)).toFixed(0)} MB`
+}
+
+export function timeAgo(dateString) {
+  if (!dateString) return 'never'
+  const seconds = Math.floor((Date.now() - new Date(dateString).getTime()) / 1000)
+  if (seconds < 0) return 'just now'
+  if (seconds < 60) return `${seconds}s ago`
+  const minutes = Math.floor(seconds / 60)
+  if (minutes < 60) return `${minutes}m ago`
+  const hours = Math.floor(minutes / 60)
+  if (hours < 24) return `${hours}h ago`
+  const days = Math.floor(hours / 24)
+  return `${days}d ago`
+}

core/http/react-ui/src/pages/NodeDetail.jsx

@@ -0,0 +1,352 @@
+import { useState, useEffect, useCallback } from 'react'
+import { useParams, useNavigate, useOutletContext } from 'react-router-dom'
+import { nodesApi } from '../utils/api'
+import PageHeader from '../components/PageHeader'
+import LoadingSpinner from '../components/LoadingSpinner'
+import ConfirmDialog from '../components/ConfirmDialog'
+import StatusPill from '../components/nodes/StatusPill'
+import CapacityEditor from '../components/nodes/CapacityEditor'
+import KeyValueChips from '../components/nodes/KeyValueChips'
+import { formatVRAM, modelStateConfig, timeAgo } from '../components/nodes/nodeStatus'
+
+// Deep-linkable node management home. Reached by clicking a roster panel on
+// /app/nodes. Surfaces what's running here plus the management affordances
+// (capacity, backends, labels, drain/resume/remove) that previously lived in
+// the expanded-row "Manage" drawer.
+export default function NodeDetail() {
+  const { id } = useParams()
+  const navigate = useNavigate()
+  const { addToast } = useOutletContext()
+  const [node, setNode] = useState(null)
+  const [models, setModels] = useState([])
+  const [backends, setBackends] = useState([])
+  const [loading, setLoading] = useState(true)
+  const [confirmRemove, setConfirmRemove] = useState(false)
+  const [confirmUnload, setConfirmUnload] = useState(null)
+  const [confirmDeleteBackend, setConfirmDeleteBackend] = useState(null)
+  // Promise-based shrink confirmation: CapacityEditor awaits this hook so the
+  // page owns the dialog (it can phrase the message with full node context).
+  const [confirmShrinkState, setConfirmShrinkState] = useState(null)
+
+  const refresh = useCallback(async () => {
+    try {
+      const n = await nodesApi.get(id)
+      setNode(n)
+      const [m, b] = await Promise.all([nodesApi.getModels(id), nodesApi.getBackends(id)])
+      setModels(Array.isArray(m) ? m : [])
+      setBackends(Array.isArray(b) ? b : [])
+    } catch (err) {
+      addToast(`Failed to load node: ${err.message}`, 'error')
+    } finally {
+      setLoading(false)
+    }
+  }, [id, addToast])
+
+  useEffect(() => { refresh() }, [refresh])
+
+  const confirmShrink = useCallback((ctx) => new Promise((resolve) => {
+    setConfirmShrinkState({ ...ctx, resolve })
+  }), [])
+
+  if (loading) return <div className="page page--wide" style={{ display: 'flex', justifyContent: 'center', padding: 'var(--spacing-xl)' }}><LoadingSpinner size="lg" /></div>
+  if (!node) return <div className="page page--wide"><PageHeader title="Node not found" /></div>
+
+  const drain = async () => { try { await nodesApi.drain(id); addToast('Node set to draining', 'success'); refresh() } catch (e) { addToast(e.message, 'error') } }
+  const resume = async () => { try { await nodesApi.resume(id); addToast('Node resumed', 'success'); refresh() } catch (e) { addToast(e.message, 'error') } }
+  const remove = async () => { try { await nodesApi.delete(id); addToast('Node removed', 'success'); navigate('/app/nodes') } catch (e) { addToast(e.message, 'error') } }
+  const unload = async (name) => { try { await nodesApi.unloadModel(id, name); addToast(`Model "${name}" unloaded`, 'success'); refresh() } catch (e) { addToast(e.message, 'error') } }
+  const upgradeBackend = async (name) => { try { await nodesApi.installBackend(id, name); addToast(`Backend "${name}" upgraded`, 'success'); refresh() } catch (e) { addToast(e.message, 'error') } }
+  const deleteBackend = async (name) => { try { await nodesApi.deleteBackend(id, name); addToast(`Backend "${name}" deleted`, 'success'); refresh() } catch (e) { addToast(e.message, 'error') } }
+  const addLabel = async (k, v) => { try { await nodesApi.mergeLabels(id, { [k]: v }); refresh() } catch (e) { addToast(e.message, 'error') } }
+  const delLabel = async (k) => { try { await nodesApi.deleteLabel(id, k); refresh() } catch (e) { addToast(e.message, 'error') } }
+
+  const usedVRAM = node.total_vram && node.available_vram != null ? node.total_vram - node.available_vram : 0
+  // {modelName: replicaCount} of loaded models so the shrink confirm can warn
+  // if the new cap is below the actual count of any single model on this node.
+  const loadedModelCounts = (() => {
+    const counts = {}
+    models.forEach(m => { if (m.state === 'loaded') counts[m.model_name] = (counts[m.model_name] || 0) + 1 })
+    return counts
+  })()
+
+  return (
+    <div className="page page--wide">
+      <PageHeader
+        eyebrow={<a onClick={() => navigate('/app/nodes')} style={{ cursor: 'pointer', color: 'var(--color-primary)' }}><i className="fas fa-arrow-left" style={{ marginRight: 6 }} aria-hidden="true" />Cluster</a>}
+        title={<><StatusPill status={node.status} /> {node.name}</>}
+        supporting={node.address}
+        actions={
+          <>
+            {node.status === 'draining'
+              ? <button className="btn btn-secondary btn-sm" onClick={resume}><i className="fas fa-play" /> Resume</button>
+              : <button className="btn btn-secondary btn-sm" onClick={drain}><i className="fas fa-pause" /> Drain</button>}
+            <button className="btn btn-danger btn-sm" onClick={() => setConfirmRemove(true)}><i className="fas fa-trash" /> Remove</button>
+          </>
+        }
+      />
+
+      {/* Inline metrics row: VRAM / in-flight - no boxes, just labelled values. */}
+      <div className="node-detail__metrics">
+        {node.total_vram > 0 && (
+          <div>
+            <div className="drawer-eyebrow">VRAM</div>
+            <span className="cell-mono">{formatVRAM(usedVRAM) || '0'} / {formatVRAM(node.total_vram)}</span>
+          </div>
+        )}
+        <div>
+          <div className="drawer-eyebrow">In-flight</div>
+          <span className="cell-mono">{node.in_flight_count || 0}</span>
+        </div>
+        {node.node_type !== 'agent' && (
+          <div style={{ minWidth: 0 }}>
+            <div className="drawer-eyebrow">Capacity</div>
+            <CapacityEditor
+              node={node}
+              loadedModelCounts={loadedModelCounts}
+              confirmShrink={confirmShrink}
+              addToast={addToast}
+              onUpdate={() => refresh()}
+            />
+          </div>
+        )}
+      </div>
+
+      {/* Running models */}
+      <div style={{ marginTop: 'var(--spacing-lg)' }}>
+        <div className="drawer-eyebrow">Running models</div>
+        {models.length === 0 ? (
+          <p style={{ fontSize: '0.8125rem', color: 'var(--color-text-muted)', margin: '0 0 var(--spacing-md) 0' }}>
+            <i className="fas fa-cube" style={{ marginRight: 6, opacity: 0.6 }} aria-hidden="true" />
+            No models loaded yet - they'll appear here when scheduled to this node.
+          </p>
+        ) : (
+          <table className="table" style={{ margin: 0 }}>
+            <thead>
+              <tr>
+                <th>Model</th>
+                <th>State</th>
+                <th>In-Flight</th>
+                <th style={{ width: 40 }}>Logs</th>
+                <th style={{ textAlign: 'right' }}>Actions</th>
+              </tr>
+            </thead>
+            <tbody>
+              {(() => {
+                // Pre-compute per-model replica counts so the disambiguation
+                // pill only renders when this node actually hosts >1 replica
+                // of the same model. Single-replica deployments stay clean.
+                const replicaCounts = {}
+                models.forEach(m => { replicaCounts[m.model_name] = (replicaCounts[m.model_name] || 0) + 1 })
+                return models.map(m => {
+                  const stCfg = modelStateConfig[m.state] || modelStateConfig.idle
+                  const showReplica = (replicaCounts[m.model_name] || 0) > 1
+                  // Per-replica process key - what the worker stores logs under and what the
+                  // store's GetLines/Subscribe match on for replica-scoped filtering.
+                  const processKey = `${m.model_name}#${m.replica_index ?? 0}`
+                  return (
+                    <tr key={m.id || `${m.model_name}#${m.replica_index ?? 0}`}>
+                      <td style={{ fontFamily: 'var(--font-mono)', fontSize: '0.8125rem' }}>
+                        {m.model_name}
+                        {showReplica && (
+                          <span
+                            className="cell-mono"
+                            aria-label={`replica ${m.replica_index ?? 0}`}
+                            title={`Replica ${m.replica_index ?? 0} on this node`}
+                            style={{
+                              marginLeft: 8, padding: '1px 6px', borderRadius: 'var(--radius-sm)',
+                              background: 'var(--color-bg-tertiary)',
+                              border: '1px solid var(--color-border-subtle)',
+                              fontSize: '0.6875rem', fontWeight: 500,
+                              color: 'var(--color-text-secondary)',
+                            }}
+                          >
+                            rep {m.replica_index ?? 0}
+                          </span>
+                        )}
+                      </td>
+                      <td>
+                        <span style={{
+                          display: 'inline-block', padding: '2px 8px', borderRadius: 'var(--radius-sm)',
+                          fontSize: '0.75rem', fontWeight: 500,
+                          background: stCfg.bg, color: stCfg.color, border: `1px solid ${stCfg.border}`,
+                        }}>
+                          {m.state}
+                        </span>
+                      </td>
+                      <td style={{ fontFamily: 'var(--font-mono)', fontSize: '0.8125rem' }}>
+                        {m.in_flight ?? 0}
+                      </td>
+                      <td>
+                        <a
+                          href="#"
+                          onClick={(e) => {
+                            e.preventDefault()
+                            // Send the replica-scoped process key (modelName#replicaIndex).
+                            navigate(`/app/node-backend-logs/${id}/${encodeURIComponent(processKey)}`)
+                          }}
+                          style={{ fontSize: '0.75rem', color: 'var(--color-primary)' }}
+                          title={showReplica ? `View backend logs for replica ${m.replica_index ?? 0}` : 'View backend logs'}
+                        >
+                          <i className="fas fa-terminal" />
+                        </a>
+                      </td>
+                      <td style={{ textAlign: 'right' }}>
+                        <button
+                          className="btn btn-danger btn-sm"
+                          title={m.in_flight > 0 ? 'Unload model (has in-flight requests)' : 'Unload model'}
+                          onClick={() => setConfirmUnload({ modelName: m.model_name, inFlight: m.in_flight ?? 0 })}
+                        >
+                          <i className="fas fa-stop" />
+                        </button>
+                      </td>
+                    </tr>
+                  )
+                })
+              })()}
+            </tbody>
+          </table>
+        )}
+      </div>
+
+      {/* Installed backends */}
+      <div style={{ marginTop: 'var(--spacing-lg)' }}>
+        <div style={{
+          display: 'flex', alignItems: 'center', justifyContent: 'space-between',
+          marginBottom: 'var(--spacing-sm)',
+        }}>
+          <div className="drawer-eyebrow" style={{ margin: 0 }}>Installed backends</div>
+          <button
+            type="button"
+            className="btn btn-secondary btn-sm"
+            onClick={() => navigate(`/app/backends?target=${encodeURIComponent(id)}`)}
+            title={`Install a backend on ${node.name}`}
+          >
+            <i className="fas fa-plus" /> Add backend
+          </button>
+        </div>
+        {backends.length === 0 ? (
+          <p style={{ fontSize: '0.8125rem', color: 'var(--color-text-muted)', margin: 0 }}>
+            None installed. <a href="#" style={{ color: 'var(--color-primary)' }} onClick={(e) => { e.preventDefault(); navigate(`/app/backends?target=${encodeURIComponent(id)}`) }}>Install one from the gallery</a> to schedule models here.
+          </p>
+        ) : (
+          <table className="table" style={{ margin: 0 }}>
+            <thead>
+              <tr>
+                <th>Name</th>
+                <th>Type</th>
+                <th>Installed At</th>
+                <th style={{ textAlign: 'right' }}>Actions</th>
+              </tr>
+            </thead>
+            <tbody>
+              {backends.map(b => (
+                <tr key={b.name}>
+                  <td style={{ fontFamily: 'var(--font-mono)', fontSize: '0.8125rem' }}>
+                    {b.name}
+                  </td>
+                  <td>
+                    <span style={{
+                      display: 'inline-block', padding: '2px 8px', borderRadius: 'var(--radius-sm)',
+                      fontSize: '0.75rem', fontWeight: 500,
+                      background: b.is_system ? 'var(--color-bg-tertiary)' : 'var(--color-primary-light)',
+                      color: b.is_system ? 'var(--color-text-muted)' : 'var(--color-primary)',
+                      border: `1px solid ${b.is_system ? 'var(--color-border-subtle)' : 'var(--color-primary-border)'}`,
+                    }}>
+                      {b.is_system ? 'system' : 'gallery'}
+                    </span>
+                  </td>
+                  <td style={{ fontSize: '0.8125rem', color: 'var(--color-text-muted)' }}>
+                    {b.installed_at ? timeAgo(b.installed_at) : '-'}
+                  </td>
+                  <td style={{ textAlign: 'right' }}>
+                    {!b.is_system && (
+                      <div style={{ display: 'inline-flex', gap: 'var(--spacing-xs)' }}>
+                        <button
+                          className="btn btn-secondary btn-sm"
+                          onClick={() => upgradeBackend(b.name)}
+                          title="Upgrade backend on this node"
+                        >
+                          <i className="fas fa-arrow-up" />
+                        </button>
+                        <button
+                          className="btn btn-danger-ghost btn-sm"
+                          onClick={() => setConfirmDeleteBackend({ backend: b.name })}
+                          title="Delete backend from this node"
+                        >
+                          <i className="fas fa-trash" />
+                        </button>
+                      </div>
+                    )}
+                  </td>
+                </tr>
+              ))}
+            </tbody>
+          </table>
+        )}
+      </div>
+
+      {/* Labels - node.replica-slots is filtered out so the Capacity editor
+          stays the single source of truth for that label. */}
+      <div style={{ marginTop: 'var(--spacing-lg)' }}>
+        <div className="drawer-eyebrow">Labels</div>
+        <KeyValueChips
+          pairs={Object.fromEntries(Object.entries(node.labels || {}).filter(([k]) => k !== 'node.replica-slots'))}
+          onAdd={addLabel}
+          onRemove={delLabel}
+          placeholderKey="key"
+          placeholderValue="value"
+          ariaLabel="Node labels"
+        />
+      </div>
+
+      <ConfirmDialog
+        open={confirmRemove}
+        title="Remove node"
+        message={`Remove "${node.name}" from the cluster? This will deregister it.`}
+        confirmLabel="Remove"
+        danger
+        onConfirm={() => { remove(); setConfirmRemove(false) }}
+        onCancel={() => setConfirmRemove(false)}
+      />
+
+      <ConfirmDialog
+        open={!!confirmUnload}
+        title="Unload Model"
+        message={
+          confirmUnload
+            ? confirmUnload.inFlight > 0
+              ? `"${confirmUnload.modelName}" currently has ${confirmUnload.inFlight} in-flight request(s). Unloading will interrupt them. Continue?`
+              : `Unload "${confirmUnload.modelName}" from ${node.name}?`
+            : ''
+        }
+        confirmLabel="Unload"
+        danger={confirmUnload?.inFlight > 0}
+        onConfirm={() => { if (confirmUnload) unload(confirmUnload.modelName); setConfirmUnload(null) }}
+        onCancel={() => setConfirmUnload(null)}
+      />
+
+      <ConfirmDialog
+        open={!!confirmDeleteBackend}
+        title="Delete Backend"
+        message={confirmDeleteBackend ? `Delete "${confirmDeleteBackend.backend}" from ${node.name}? This removes the backend files from this node only.` : ''}
+        confirmLabel="Delete"
+        danger
+        onConfirm={() => { if (confirmDeleteBackend) deleteBackend(confirmDeleteBackend.backend); setConfirmDeleteBackend(null) }}
+        onCancel={() => setConfirmDeleteBackend(null)}
+      />
+
+      <ConfirmDialog
+        open={!!confirmShrinkState}
+        title="Reduce replica capacity"
+        message={
+          confirmShrinkState
+            ? `${node.name} currently has ${confirmShrinkState.currentLoaded} replica(s) of at least one model loaded. Reducing the cap to ${confirmShrinkState.newValue} won't evict anything immediately - running replicas keep going, but the reconciler will trim down on the next idle window. Continue?`
+            : ''
+        }
+        confirmLabel="Reduce"
+        onConfirm={() => { confirmShrinkState?.resolve(true); setConfirmShrinkState(null) }}
+        onCancel={() => { confirmShrinkState?.resolve(false); setConfirmShrinkState(null) }}
+      />
+    </div>
+  )
+}

core/http/react-ui/src/pages/Scheduling.jsx

@@ -0,0 +1,438 @@
+import { useState, useEffect, useCallback } from 'react'
+import { useOutletContext } from 'react-router-dom'
+import { useTranslation } from 'react-i18next'
+import { nodesApi } from '../utils/api'
+import PageHeader from '../components/PageHeader'
+import ConfirmDialog from '../components/ConfirmDialog'
+import ResponsiveTable from '../components/ResponsiveTable'
+import SearchableModelSelect from '../components/SearchableModelSelect'
+import KeyValueChips from '../components/nodes/KeyValueChips'
+
+// Numeric input with quick-pick preset chips. Picked over a slider because
+// replica counts are exact specs (operator math), not fuzzy estimates. The
+// chips give one-click access to common values without the slider's
+// precision/special-value problems (e.g. MaxReplicas=0 = "no limit").
+function ReplicaInput({ id, label, value, onChange, presets }) {
+  return (
+    <div style={{ flex: 1 }}>
+      <label className="form-label" htmlFor={id}>{label}</label>
+      <input
+        id={id}
+        className="input"
+        type="number"
+        min={0}
+        value={value}
+        onChange={e => onChange(parseInt(e.target.value) || 0)}
+      />
+      <div style={{ display: 'flex', gap: 4, flexWrap: 'wrap', marginTop: 6 }}>
+        {presets.map(({ v, l }) => {
+          const active = value === v
+          return (
+            <button
+              key={v}
+              type="button"
+              onClick={() => onChange(v)}
+              aria-pressed={active}
+              className="cell-mono"
+              style={{
+                padding: '2px 8px',
+                borderRadius: 'var(--radius-sm)',
+                fontSize: '0.6875rem',
+                fontWeight: 500,
+                cursor: 'pointer',
+                background: active ? 'var(--color-primary-light)' : 'transparent',
+                border: `1px solid ${active ? 'var(--color-primary-border)' : 'var(--color-border-subtle)'}`,
+                color: active ? 'var(--color-primary)' : 'var(--color-text-muted)',
+              }}
+            >{l || v}</button>
+          )
+        })}
+      </div>
+    </div>
+  )
+}
+
+function SchedulingForm({ onSave, onCancel }) {
+  const [mode, setMode] = useState('placement')
+  const [modelName, setModelName] = useState('')
+  // Selector is now a chip-builder map instead of a comma-separated string.
+  // Operators were copying syntax from docs and missing commas; the chip UI
+  // makes the key=value structure self-documenting.
+  const [selector, setSelector] = useState({})
+  const [minReplicas, setMinReplicas] = useState(1)
+  const [maxReplicas, setMaxReplicas] = useState(0)
+  // Prefix-cache routing controls. Empty routePolicy means "inherit the
+  // cluster default"; the three thresholds at 0 likewise inherit, so they
+  // stay out of the POST body's effective override only when explicitly set.
+  const [routePolicy, setRoutePolicy] = useState('')
+  const [balanceAbsThreshold, setBalanceAbsThreshold] = useState(0)
+  const [balanceRelThreshold, setBalanceRelThreshold] = useState(0)
+  const [minPrefixMatch, setMinPrefixMatch] = useState(0)
+
+  const hasSelector = Object.keys(selector).length > 0
+
+  const isValid = () => {
+    if (!modelName) return false
+    if (mode === 'placement') return hasSelector
+    if (mode === 'spread') return true
+    return minReplicas > 0 || maxReplicas > 0
+  }
+
+  const handleSubmit = () => {
+    onSave({
+      model_name: modelName,
+      node_selector: hasSelector ? selector : undefined,
+      min_replicas: mode === 'autoscaling' ? minReplicas : 0,
+      max_replicas: mode === 'autoscaling' ? maxReplicas : 0,
+      spread_all: mode === 'spread',
+      route_policy: routePolicy,
+      balance_abs_threshold: balanceAbsThreshold,
+      balance_rel_threshold: balanceRelThreshold,
+      min_prefix_match: minPrefixMatch,
+    })
+  }
+
+  return (
+    <div className="card" style={{ padding: 'var(--spacing-lg)', marginBottom: 'var(--spacing-md)' }}>
+      {/* Mode selector — uses the project's segmented control instead of two
+          50%-width filled buttons that competed visually with the actual
+          primary action (Save). */}
+      <div role="radiogroup" aria-label="Scheduling mode" className="segmented" style={{ marginBottom: 'var(--spacing-xs)' }}>
+        <button
+          type="button" role="radio" aria-checked={mode === 'placement'}
+          className={`segmented__item${mode === 'placement' ? ' is-active' : ''}`}
+          onClick={() => setMode('placement')}
+        >
+          <i className="fas fa-thumbtack" aria-hidden="true" /> Pin to nodes
+        </button>
+        <button
+          type="button" role="radio" aria-checked={mode === 'autoscaling'}
+          className={`segmented__item${mode === 'autoscaling' ? ' is-active' : ''}`}
+          onClick={() => setMode('autoscaling')}
+        >
+          <i className="fas fa-arrows-up-down" aria-hidden="true" /> Auto-scale
+        </button>
+        <button
+          type="button" role="radio" aria-checked={mode === 'spread'}
+          className={`segmented__item${mode === 'spread' ? ' is-active' : ''}`}
+          onClick={() => setMode('spread')}
+        >
+          <i className="fas fa-network-wired" aria-hidden="true" /> Spread to all
+        </button>
+      </div>
+      <p style={{ fontSize: '0.8125rem', color: 'var(--color-text-muted)', margin: '0 0 var(--spacing-lg) 0' }}>
+        {mode === 'placement'
+          ? 'Restrict this model to specific nodes. Loaded on demand, evictable when idle.'
+          : mode === 'spread'
+          ? 'Run one replica on every node matching the selector (all healthy nodes when empty). Tracks nodes joining and leaving.'
+          : 'Maintain a target replica count across the cluster. Min ≥ 1 protects from eviction.'}
+      </p>
+
+      {/* Linear vertical flow — model picker is the visual focus, then the
+          mode-specific fields below. No 2-column grid (the mismatched widths
+          made the form look raw). */}
+      <div style={{ display: 'flex', flexDirection: 'column', gap: 'var(--spacing-md)' }}>
+        <div>
+          <label className="form-label" htmlFor="sched-model">Model</label>
+          {/* Searchable combobox so a long gallery doesn't force the operator
+              to scroll through hundreds of entries. Free-text is allowed —
+              you can pre-create a rule for a model that hasn't been
+              installed yet, which is a real workflow when standing up a new
+              node and pre-staging its scheduling policy. */}
+          <SearchableModelSelect
+            value={modelName}
+            onChange={setModelName}
+            placeholder="Type to search models, or paste a name..."
+          />
+        </div>
+
+        <div>
+          <label className="form-label">
+            Node selector{mode === 'placement' ? '' : ' (optional)'}
+          </label>
+          <KeyValueChips
+            pairs={selector}
+            onAdd={(k, v) => setSelector(prev => ({ ...prev, [k]: v }))}
+            onRemove={(k) => setSelector(prev => { const n = { ...prev }; delete n[k]; return n })}
+            placeholderKey="key (e.g. gpu.vendor)"
+            placeholderValue="value (e.g. nvidia)"
+            ariaLabel="Node selector"
+          />
+          <span style={{ fontSize: '0.75rem', color: 'var(--color-text-muted)', display: 'block', marginTop: 6 }}>
+            {mode === 'placement'
+              ? 'Models will load only on nodes that match all listed labels.'
+              : (hasSelector ? 'Replicas land only on matching nodes.' : 'Empty = any healthy node.')}
+          </span>
+        </div>
+
+        {mode === 'autoscaling' && (
+          <div style={{ display: 'flex', gap: 'var(--spacing-md)' }}>
+            <ReplicaInput
+              id="sched-min"
+              label="Min replicas"
+              value={minReplicas}
+              onChange={setMinReplicas}
+              presets={[{ v: 1 }, { v: 2 }, { v: 3 }, { v: 4 }]}
+            />
+            <ReplicaInput
+              id="sched-max"
+              label="Max replicas"
+              value={maxReplicas}
+              onChange={setMaxReplicas}
+              presets={[{ v: 0, l: 'no limit' }, { v: 2 }, { v: 4 }, { v: 8 }]}
+            />
+          </div>
+        )}
+
+        {/* Per-model routing policy. Left empty/zero these inherit the
+            cluster-wide defaults; set them to override how requests for this
+            model are spread across replicas. */}
+        <div>
+          <label className="form-label" htmlFor="sched-route-policy">Routing policy</label>
+          <select
+            id="sched-route-policy"
+            className="input"
+            value={routePolicy}
+            onChange={e => setRoutePolicy(e.target.value)}
+          >
+            <option value="">Default (cluster setting)</option>
+            <option value="round_robin">Round Robin</option>
+            <option value="prefix_cache">Prefix Cache</option>
+          </select>
+          <span style={{ fontSize: '0.75rem', color: 'var(--color-text-muted)', display: 'block', marginTop: 6 }}>
+            Prefix Cache routes shared-prefix requests to the same replica to reuse its KV cache, falling back to round-robin when replicas are imbalanced.
+          </span>
+        </div>
+
+        {routePolicy === 'prefix_cache' && (
+          <div style={{ display: 'flex', gap: 'var(--spacing-md)' }}>
+            <div style={{ flex: 1 }}>
+              <label className="form-label" htmlFor="sched-min-prefix-match">Min prefix match</label>
+              <input
+                id="sched-min-prefix-match"
+                className="input"
+                type="number"
+                step="0.05"
+                min="0"
+                max="1"
+                value={minPrefixMatch}
+                onChange={e => setMinPrefixMatch(parseFloat(e.target.value) || 0)}
+              />
+              <span style={{ fontSize: '0.75rem', color: 'var(--color-text-muted)', display: 'block', marginTop: 6 }}>
+                Fraction of the prompt (0..1) that must match a cached prefix before affinity kicks in. 0 inherits the default.
+              </span>
+            </div>
+            <div style={{ flex: 1 }}>
+              <label className="form-label" htmlFor="sched-balance-abs">Balance abs threshold</label>
+              <input
+                id="sched-balance-abs"
+                className="input"
+                type="number"
+                min="0"
+                value={balanceAbsThreshold}
+                onChange={e => setBalanceAbsThreshold(parseInt(e.target.value) || 0)}
+              />
+              <span style={{ fontSize: '0.75rem', color: 'var(--color-text-muted)', display: 'block', marginTop: 6 }}>
+                Max absolute in-flight gap allowed before falling back to round-robin. 0 inherits the default.
+              </span>
+            </div>
+            <div style={{ flex: 1 }}>
+              <label className="form-label" htmlFor="sched-balance-rel">Balance rel threshold</label>
+              <input
+                id="sched-balance-rel"
+                className="input"
+                type="number"
+                step="0.1"
+                min="0"
+                value={balanceRelThreshold}
+                onChange={e => setBalanceRelThreshold(parseFloat(e.target.value) || 0)}
+              />
+              <span style={{ fontSize: '0.75rem', color: 'var(--color-text-muted)', display: 'block', marginTop: 6 }}>
+                Max relative in-flight ratio (&gt;= 1) allowed before falling back to round-robin. 0 inherits the default.
+              </span>
+            </div>
+          </div>
+        )}
+      </div>
+
+      {/* Hairline divider above the actions, matching the project's form pattern. */}
+      <div style={{
+        display: 'flex', gap: 'var(--spacing-sm)', justifyContent: 'flex-end',
+        marginTop: 'var(--spacing-lg)', paddingTop: 'var(--spacing-md)',
+        borderTop: '1px solid var(--color-border-subtle)',
+      }}>
+        <button className="btn btn-secondary btn-sm" onClick={onCancel}>Cancel</button>
+        <button className="btn btn-primary btn-sm" onClick={handleSubmit} disabled={!isValid()}>Save rule</button>
+      </div>
+    </div>
+  )
+}
+
+export default function Scheduling() {
+  const { addToast } = useOutletContext()
+  const { t } = useTranslation('admin')
+  const [schedulingConfigs, setSchedulingConfigs] = useState([])
+  const [showForm, setShowForm] = useState(false)
+  const [confirmDelete, setConfirmDelete] = useState(null)
+
+  const fetchScheduling = useCallback(async () => {
+    try {
+      const data = await nodesApi.listScheduling()
+      setSchedulingConfigs(Array.isArray(data) ? data : [])
+    } catch { setSchedulingConfigs([]) }
+  }, [])
+
+  useEffect(() => { fetchScheduling() }, [fetchScheduling])
+
+  const handleSave = async (config) => {
+    try {
+      await nodesApi.setScheduling(config)
+      addToast('Scheduling rule saved', 'success')
+      setShowForm(false)
+      fetchScheduling()
+    } catch (err) { addToast(`Failed to save rule: ${err.message}`, 'error') }
+  }
+
+  const handleDelete = async (model) => {
+    try {
+      await nodesApi.deleteScheduling(model)
+      addToast('Scheduling rule removed', 'success')
+      setConfirmDelete(null)
+      fetchScheduling()
+    } catch (err) { addToast(`Failed to remove rule: ${err.message}`, 'error') }
+  }
+
+  return (
+    <div className="page page--wide">
+      <PageHeader
+        title={<><i className="fas fa-calendar-alt" style={{ marginRight: 'var(--spacing-sm)' }} />{t('scheduling.title')}</>}
+        supporting={t('scheduling.subtitle')}
+      />
+      <div>
+        <button className="btn btn-primary btn-sm" style={{ marginBottom: 'var(--spacing-md)' }}
+          onClick={() => setShowForm(f => !f)}>
+          <i className="fas fa-plus" style={{ marginRight: 6 }} />
+          Add Scheduling Rule
+        </button>
+        {showForm && <SchedulingForm onSave={handleSave} onCancel={() => setShowForm(false)} />}
+        {schedulingConfigs.length === 0 && !showForm ? (
+          <p style={{ fontSize: '0.875rem', color: 'var(--color-text-muted)', textAlign: 'center', padding: 'var(--spacing-xl) 0' }}>
+            No scheduling rules configured. Add a rule to control how models are placed on nodes.
+          </p>
+        ) : schedulingConfigs.length > 0 && (
+          <ResponsiveTable>
+              <thead><tr>
+                <th>Model</th>
+                <th>Mode</th>
+                <th>Node Selector</th>
+                <th>Min Replicas</th>
+                <th>Max Replicas</th>
+                <th>Routing</th>
+                <th>Thresholds</th>
+                <th>Status</th>
+                <th style={{ textAlign: 'right' }}>Actions</th>
+              </tr></thead>
+              <tbody>
+                {schedulingConfigs.map(cfg => {
+                  const isSpread = !!cfg.spread_all
+                  const isAutoScaling = !isSpread && (cfg.min_replicas > 0 || cfg.max_replicas > 0)
+                  const hasSelector = !!cfg.node_selector
+                  const modeLabel = isSpread ? 'Spread' : isAutoScaling ? 'Auto-scaling' : hasSelector ? 'Placement' : 'Inactive'
+                  const modeColor = isSpread ? 'var(--color-warning)' : isAutoScaling ? 'var(--color-success)' : hasSelector ? 'var(--color-primary)' : 'var(--color-text-muted)'
+                  // Cooldown: reconciler tripped the circuit breaker because cluster
+                  // capacity is exhausted. Surface so the operator sees it instead
+                  // of the model silently failing to scale.
+                  const unsatisfiableUntil = cfg.unsatisfiable_until ? new Date(cfg.unsatisfiable_until) : null
+                  const isUnsatisfiable = unsatisfiableUntil && unsatisfiableUntil.getTime() > Date.now()
+                  return (
+                  <tr key={cfg.id || cfg.model_name}>
+                    <td style={{ fontWeight: 600, fontSize: '0.875rem' }}>{cfg.model_name}</td>
+                    <td>
+                      <span style={{
+                        display: 'inline-block', fontSize: '0.75rem', padding: '2px 8px', borderRadius: "var(--radius-sm)",
+                        background: 'var(--color-bg-tertiary)', border: `1px solid ${modeColor}`,
+                        color: modeColor, fontWeight: 600,
+                      }}>{modeLabel}</span>
+                    </td>
+                    <td>
+                      {cfg.node_selector ? (() => {
+                        try {
+                          const sel = typeof cfg.node_selector === 'string' ? JSON.parse(cfg.node_selector) : cfg.node_selector
+                          return Object.entries(sel).map(([k,v]) => (
+                            <span key={k} style={{
+                              display: 'inline-block', fontSize: '0.75rem', padding: '2px 6px', borderRadius: "var(--radius-sm)",
+                              background: 'var(--color-bg-tertiary)', border: '1px solid var(--color-border-subtle)',
+                              fontFamily: 'var(--font-mono)', marginRight: 4,
+                            }}>{k}={v}</span>
+                          ))
+                        } catch { return <span style={{ color: 'var(--color-text-muted)', fontSize: '0.8125rem' }}>{cfg.node_selector}</span> }
+                      })() : <span style={{ color: 'var(--color-text-muted)', fontSize: '0.8125rem' }}>Any node</span>}
+                    </td>
+                    <td style={{ fontFamily: 'var(--font-mono)' }}>
+                      {isSpread
+                        ? <span style={{
+                            display: 'inline-block', fontSize: '0.75rem', padding: '2px 8px', borderRadius: "var(--radius-sm)",
+                            background: 'var(--color-bg-tertiary)', border: '1px solid var(--color-warning)',
+                            color: 'var(--color-warning)', fontWeight: 600, fontFamily: 'var(--font-sans)',
+                          }}>Spread: all matching nodes</span>
+                        : isAutoScaling ? cfg.min_replicas : '-'}
+                    </td>
+                    <td style={{ fontFamily: 'var(--font-mono)' }}>
+                      {isSpread ? '-' : isAutoScaling ? (cfg.max_replicas || 'no limit') : '-'}
+                    </td>
+                    <td style={{ fontSize: '0.8125rem' }}>
+                      {cfg.route_policy || 'default'}
+                    </td>
+                    <td style={{ fontFamily: 'var(--font-mono)', fontSize: '0.75rem', color: 'var(--color-text-muted)' }}>
+                      {cfg.route_policy === 'prefix_cache' ? (
+                        <>
+                          <div>match: {cfg.min_prefix_match ? cfg.min_prefix_match : 'inherit'}</div>
+                          <div>abs: {cfg.balance_abs_threshold ? cfg.balance_abs_threshold : 'inherit'}</div>
+                          <div>rel: {cfg.balance_rel_threshold ? cfg.balance_rel_threshold : 'inherit'}</div>
+                        </>
+                      ) : '-'}
+                    </td>
+                    <td>
+                      {isUnsatisfiable ? (
+                        <span
+                          title={`Reconciler couldn't satisfy this rule (capacity exhausted). Will retry by ${unsatisfiableUntil.toLocaleString()}, or sooner on a node lifecycle change.`}
+                          style={{
+                            display: 'inline-block', fontSize: '0.75rem', padding: '2px 8px',
+                            borderRadius: 'var(--radius-sm)', fontWeight: 600,
+                            background: 'var(--color-bg-tertiary)',
+                            border: '1px solid var(--color-warning, #d97706)',
+                            color: 'var(--color-warning, #d97706)',
+                          }}
+                        >
+                          <i className="fas fa-exclamation-triangle" style={{ marginRight: 4 }} />
+                          Unsatisfiable until {unsatisfiableUntil.toLocaleTimeString([], { hour: '2-digit', minute: '2-digit' })}
+                        </span>
+                      ) : (
+                        <span style={{ fontSize: '0.8125rem', color: 'var(--color-text-muted)' }}>OK</span>
+                      )}
+                    </td>
+                    <td style={{ textAlign: 'right' }}>
+                      <button className="btn btn-danger btn-sm" onClick={() => setConfirmDelete(cfg.model_name)}>
+                        <i className="fas fa-trash" />
+                      </button>
+                    </td>
+                  </tr>
+                  )
+                })}
+              </tbody>
+          </ResponsiveTable>
+        )}
+      </div>
+
+      <ConfirmDialog
+        open={!!confirmDelete}
+        title="Remove scheduling rule"
+        message={confirmDelete ? `Remove the scheduling rule for "${confirmDelete}"?` : ''}
+        confirmLabel="Remove"
+        danger
+        onConfirm={() => confirmDelete && handleDelete(confirmDelete)}
+        onCancel={() => setConfirmDelete(null)}
+      />
+    </div>
+  )
+}

core/http/react-ui/src/router.jsx

@@ -69,7 +69,9 @@ const Studio = page('studio', () => import('./pages/Studio'))
 const FaceRecognition = page('face', () => import('./pages/FaceRecognition'))
 const VoiceRecognition = page('voice', () => import('./pages/VoiceRecognition'))
 const Nodes = page('nodes', () => import('./pages/Nodes'))
+const Scheduling = page('scheduling', () => import('./pages/Scheduling'))
 const NodeBackendLogs = page(null, () => import('./pages/NodeBackendLogs'))
+const NodeDetail = page(null, () => import('./pages/NodeDetail'))
 const NotFound = page(null, () => import('./pages/NotFound'))
 const Usage = page('usage', () => import('./pages/Usage'))
 const Users = page('users', () => import('./pages/Users'))
@@ -152,6 +154,8 @@ const appChildren = [
       { path: 'backend-logs/:modelId', element: <Admin><BackendLogs /></Admin> },
       { path: 'p2p', element: <Admin><P2P /></Admin> },
       { path: 'nodes', element: <Admin><Nodes /></Admin> },
+      { path: 'nodes/:id', element: <Admin><NodeDetail /></Admin> },
+      { path: 'scheduling', element: <Admin><Scheduling /></Admin> },
       { path: 'node-backend-logs/:nodeId/:modelId', element: <Admin><NodeBackendLogs /></Admin> },
       { path: 'usage', element: <Usage /> },
       { path: 'users', element: <RequireAuthEnabled><Admin><Users /></Admin></RequireAuthEnabled> },

core/http/react-ui/src/utils/api.js

@@ -568,6 +568,7 @@ export const nodesApi = {
     method: 'DELETE',
   }),
   listScheduling: () => fetchJSON(API_CONFIG.endpoints.nodesScheduling),
+  allModels: () => fetchJSON(API_CONFIG.endpoints.nodesModels),
   setScheduling: (config) => postJSON(API_CONFIG.endpoints.nodesScheduling, config),
   deleteScheduling: (model) => fetchJSON(API_CONFIG.endpoints.nodesSchedulingModel(model), { method: 'DELETE' }),
 }

core/http/react-ui/src/utils/config.js

@@ -144,6 +144,7 @@ export const API_CONFIG = {
     nodeLabelKey: (id, key) => `/api/nodes/${id}/labels/${key}`,
     nodeMaxReplicasPerModel: (id) => `/api/nodes/${id}/max-replicas-per-model`,
     nodesScheduling: '/api/nodes/scheduling',
+    nodesModels: '/api/nodes/models',
     nodesSchedulingModel: (model) => `/api/nodes/scheduling/${encodeURIComponent(model)}`,
   },
 }

core/http/routes/nodes.go

@@ -71,6 +71,9 @@ func RegisterNodeAdminRoutes(e *echo.Echo, registry *nodes.NodeRegistry, unloade
 	admin := e.Group("/api/nodes", readyMw, adminMw)
 	admin.GET("", localai.ListNodesEndpoint(registry))
 
+	// Cluster-wide loaded models (registered before /:id to avoid route conflicts)
+	admin.GET("/models", localai.ListAllNodeModelsEndpoint(registry))
+
 	// Model scheduling (registered before /:id to avoid route conflicts)
 	admin.GET("/scheduling", localai.ListSchedulingEndpoint(registry))
 	admin.GET("/scheduling/:model", localai.GetSchedulingEndpoint(registry))

core/services/distributed/gallery.go

@@ -79,21 +79,29 @@ func (s *GalleryStore) Create(op *GalleryOperationRecord) error {
 	}).Create(op).Error
 }
 
-// UpdateProgress updates progress for an operation.
-func (s *GalleryStore) UpdateProgress(id string, progress float64, message, downloadedSize string) error {
+// UpdateProgress updates progress for an operation. The cancellable flag is
+// persisted on every tick so a replica that restarts mid-install rehydrates the
+// op as still cancellable — otherwise the column keeps its Create-time zero
+// value (false), the UI hides the cancel button, and the orphaned op can only
+// be dismissed by waiting for the 30-minute stale reaper.
+func (s *GalleryStore) UpdateProgress(id string, progress float64, message, downloadedSize string, cancellable bool) error {
 	return s.db.Model(&GalleryOperationRecord{}).Where("id = ?", id).Updates(map[string]any{
 		"progress":             progress,
 		"message":              message,
 		"downloaded_file_size": downloadedSize,
+		"cancellable":          cancellable,
 		"updated_at":           time.Now(),
 	}).Error
 }
 
-// UpdateStatus updates the status of an operation.
+// UpdateStatus updates the status of an operation. A terminal status is never
+// cancellable, so the flag is cleared here to keep the persisted row consistent
+// with what the UI should offer.
 func (s *GalleryStore) UpdateStatus(id, status, errMsg string) error {
 	updates := map[string]any{
-		"status":     status,
-		"updated_at": time.Now(),
+		"status":      status,
+		"cancellable": false,
+		"updated_at":  time.Now(),
 	}
 	if errMsg != "" {
 		updates["error"] = errMsg

core/services/galleryop/cancellable_persist_test.go

@@ -0,0 +1,56 @@
+package galleryop_test
+
+import (
+	. "github.com/onsi/ginkgo/v2"
+	. "github.com/onsi/gomega"
+
+	"github.com/mudler/LocalAI/core/config"
+	"github.com/mudler/LocalAI/core/services/distributed"
+	"github.com/mudler/LocalAI/core/services/galleryop"
+	"github.com/mudler/LocalAI/core/services/testutil"
+)
+
+// Reproduces "an in-flight install can't be cancelled after a restart". The
+// live install path marks OpStatus.Cancellable=true on every progress tick, but
+// UpdateStatus persisted progress/status to the gallery store WITHOUT the
+// cancellable flag, and Create defaulted it to false. So after a replica
+// restart Hydrate rebuilt the op with Cancellable=false, /api/operations
+// reported cancellable:false, and the UI hid the cancel button — the orphaned
+// op lingered until the 30-minute stale reaper expired it. The cancellable
+// state must be persisted so a rehydrated in-flight op stays cancellable.
+var _ = Describe("GalleryService cancellable persistence across restart", func() {
+	It("rehydrates an in-flight op as still cancellable", func() {
+		db := testutil.SetupTestDB()
+		store, err := distributed.NewGalleryStore(db)
+		Expect(err).ToNot(HaveOccurred())
+
+		svc := galleryop.NewGalleryService(&config.ApplicationConfig{}, nil)
+		svc.SetGalleryStore(store)
+
+		// Seed the in-flight op row as the worker goroutine does on admission.
+		Expect(store.Create(&distributed.GalleryOperationRecord{
+			ID:                 "op-inflight",
+			GalleryElementName: "llama-cpp-development",
+			OpType:             "backend_install",
+			Status:             "pending",
+		})).To(Succeed())
+
+		// Simulate a progress tick: the live path always marks installs
+		// cancellable while they are downloading/processing.
+		svc.UpdateStatus("op-inflight", &galleryop.OpStatus{
+			Message:     "downloading",
+			Progress:    25,
+			Cancellable: true,
+		})
+
+		// A fresh replica boots and hydrates from the store.
+		fresh := galleryop.NewGalleryService(&config.ApplicationConfig{}, nil)
+		fresh.SetGalleryStore(store)
+		Expect(fresh.Hydrate()).To(Succeed())
+
+		st := fresh.GetStatus("op-inflight")
+		Expect(st).ToNot(BeNil(), "the in-flight op must hydrate after a restart")
+		Expect(st.Cancellable).To(BeTrue(),
+			"a still-active install must rehydrate as cancellable so the admin can dismiss it")
+	})
+})

core/services/galleryop/service.go

@@ -167,7 +167,7 @@ func (g *GalleryService) UpdateStatus(s string, op *OpStatus) {
 				xlog.Warn("Failed to persist gallery operation status", "op_id", s, "error", err)
 			}
 		} else {
-			if err := store.UpdateProgress(s, op.Progress, op.Message, op.DownloadedFileSize); err != nil {
+			if err := store.UpdateProgress(s, op.Progress, op.Message, op.DownloadedFileSize, op.Cancellable); err != nil {
 				xlog.Warn("Failed to persist gallery operation progress", "op_id", s, "error", err)
 			}
 		}
@@ -467,6 +467,7 @@ func (g *GalleryService) Start(c context.Context, cl *config.ModelConfigLoader,
 						GalleryElementName: op.GalleryElementName,
 						OpType:             "backend_install",
 						Status:             "pending",
+						Cancellable:        true,
 					})
 				}
 				err := g.backendHandler(&op, systemState)
@@ -499,6 +500,8 @@ func (g *GalleryService) Start(c context.Context, cl *config.ModelConfigLoader,
 						GalleryElementName: op.GalleryElementName,
 						OpType:             opType,
 						Status:             "pending",
+						// A delete is not cancellable; an install is.
+						Cancellable: !op.Delete,
 					})
 				}
 				err := g.modelHandler(&op, cl, systemState)

core/services/messaging/subjects.go

@@ -64,6 +64,22 @@ func SubjectGalleryProgress(opID string) string {
 	return subjectGalleryPrefix + sanitizeSubjectToken(opID) + ".progress"
 }
 
+// SubjectStagingProgress returns the NATS subject a frontend replica publishes
+// file-staging progress on. Staging progress is otherwise per-process state
+// (the SmartRouter's in-memory StagingTracker), so without this broadcast a
+// /api/operations poll that round-robins onto a replica that did not originate
+// the staging op sees nothing - the progress row flickers in multi-replica
+// deployments. Peers subscribe to the wildcard and merge.
+func SubjectStagingProgress(modelID string) string {
+	return subjectStagingPrefix + sanitizeSubjectToken(modelID) + ".progress"
+}
+
+const subjectStagingPrefix = "staging."
+
+// SubjectStagingProgressWildcard matches every replica's staging-progress
+// broadcasts so a peer can mirror staging ops it did not originate.
+const SubjectStagingProgressWildcard = "staging.*.progress"
+
 // SubjectGalleryOpStart and SubjectGalleryOpEnd are broadcast subjects for the
 // in-memory OpCache lifecycle. Frontend replicas publish to these when an
 // admin admits a new install/delete (Start) and when an operation is

core/services/nodes/router.go

@@ -359,8 +359,21 @@ func (r *SmartRouter) Route(ctx context.Context, modelID, modelName, backendType
 		}
 	}
 
-	// Step 2: Model not loaded — schedule loading with distributed lock to prevent duplicates
-	loadModel := func() (*RouteResult, error) {
+	// Step 2: Model not loaded — schedule loading with distributed lock to prevent duplicates.
+	//
+	// Detach the cold-load from the caller's context. Staging a model can
+	// transfer multiple GB to a worker, which takes far longer than any client
+	// keeps its HTTP request open — a browser refresh, an ingress/LB idle
+	// timeout, or a round-robined retry landing on another replica all cancel
+	// the request context. If staging were bound to it, the multi-GB upload
+	// aborts with "context canceled" mid-transfer and large models can never
+	// finish staging (the model-load outage). WithoutCancel keeps the request's
+	// values (prefix chain, etc.) but drops its cancellation/deadline. Each
+	// long step still has its own bound (the file stager's resume budget,
+	// LoadModel's 5m timeout), and the per-model advisory lock below de-dupes
+	// concurrent loaders across replicas.
+	loadCtx := context.WithoutCancel(ctx)
+	loadModel := func(ctx context.Context) (*RouteResult, error) {
 		// Re-check after acquiring lock — another request may have loaded it
 		node, nm, err := r.registry.FindAndLockNodeWithModel(ctx, trackingKey, candidateNodeIDs, pref)
 		if err == nil && node != nil {
@@ -433,9 +446,9 @@ func (r *SmartRouter) Route(ctx context.Context, modelID, modelName, backendType
 	if r.db != nil {
 		lockKey := advisorylock.KeyFromString("model-load:" + trackingKey)
 		var result *RouteResult
-		lockErr := advisorylock.WithLockCtx(ctx, r.db, lockKey, func() error {
+		lockErr := advisorylock.WithLockCtx(loadCtx, r.db, lockKey, func() error {
 			var err error
-			result, err = loadModel()
+			result, err = loadModel(loadCtx)
 			return err
 		})
 		if lockErr != nil {
@@ -444,7 +457,7 @@ func (r *SmartRouter) Route(ctx context.Context, modelID, modelName, backendType
 		return result, nil
 	}
 	// No DB (non-distributed) — proceed without lock
-	return loadModel()
+	return loadModel(loadCtx)
 }
 
 // parseSelectorJSON decodes a JSON node selector string into a map.

core/services/nodes/router_staging_context_test.go

@@ -0,0 +1,80 @@
+package nodes
+
+import (
+	"context"
+	"errors"
+	"os"
+	"path/filepath"
+
+	. "github.com/onsi/ginkgo/v2"
+	. "github.com/onsi/gomega"
+
+	"github.com/mudler/LocalAI/core/services/messaging"
+	pb "github.com/mudler/LocalAI/pkg/grpc/proto"
+)
+
+// cancelOnStageStager simulates the triggering HTTP request being abandoned
+// (client disconnect, ingress idle-timeout) the moment a multi-GB file starts
+// staging. It cancels the request context and records whether the context the
+// stager itself received was cancelled as a result.
+type cancelOnStageStager struct {
+	fakeFileStager
+	cancelRequest context.CancelFunc
+	staged        bool
+	ctxErrOnStage error
+}
+
+func (s *cancelOnStageStager) EnsureRemote(ctx context.Context, _, _, key string) (string, error) {
+	s.staged = true
+	// Mid-transfer: the client gives up on the (minutes-long) request.
+	if s.cancelRequest != nil {
+		s.cancelRequest()
+	}
+	// A multi-GB upload must survive this. If staging were bound to the
+	// request context, ctx is now cancelled and the real HTTP stager would
+	// abort with "context canceled" — exactly the production outage.
+	s.ctxErrOnStage = ctx.Err()
+	return "/remote/" + key, nil
+}
+
+var _ = Describe("Route cold-load staging context", func() {
+	It("detaches staging from the request context so a client disconnect cannot abort a multi-GB transfer", func() {
+		// A real model file so stageModelFiles actually calls the stager
+		// (non-existent paths are skipped).
+		tmp := GinkgoT().TempDir()
+		modelFile := filepath.Join(tmp, "big.gguf")
+		Expect(os.WriteFile(modelFile, []byte("weights"), 0o644)).To(Succeed())
+
+		reg := &fakeModelRouter{
+			findAndLockErr: errors.New("not loaded"),
+			findIdleNode:   &BackendNode{ID: "n1", Name: "worker-1", Address: "10.0.0.1:50051"},
+		}
+		backend := &stubBackend{loadResult: &pb.Result{Success: true}}
+		factory := &stubClientFactory{client: backend}
+		unloader := &fakeUnloader{installReply: &messaging.BackendInstallReply{
+			Success: true,
+			Address: "10.0.0.1:9001",
+		}}
+		stager := &cancelOnStageStager{}
+
+		router := NewSmartRouter(reg, SmartRouterOptions{
+			Unloader:      unloader,
+			ClientFactory: factory,
+			FileStager:    stager,
+			// DB nil: no advisory lock, exercises the same detached load ctx.
+		})
+
+		ctx, cancel := context.WithCancel(context.Background())
+		stager.cancelRequest = cancel
+		defer cancel()
+
+		result, err := router.Route(ctx, "big-model", filepath.Join("models", "big.gguf"), "llama-cpp",
+			&pb.ModelOptions{Model: "big.gguf", ModelFile: modelFile}, false)
+
+		Expect(err).ToNot(HaveOccurred())
+		Expect(result).ToNot(BeNil())
+		Expect(stager.staged).To(BeTrue(), "staging must have been attempted")
+		Expect(stager.ctxErrOnStage).ToNot(HaveOccurred(),
+			"staging context must survive cancellation of the triggering request")
+	})
+})

core/services/nodes/staging_progress.go

@@ -5,58 +5,138 @@ import (
 	"fmt"
 	"sync"
 	"time"
+
+	"github.com/mudler/LocalAI/core/services/messaging"
 )
 
 // StagingStatus represents the current progress of a model staging operation.
 type StagingStatus struct {
-	ModelID    string  `json:"model_id"`
-	NodeName   string  `json:"node_name"`
-	FileName   string  `json:"file_name"`
-	BytesSent  int64   `json:"bytes_sent"`
-	TotalBytes int64   `json:"total_bytes"`
-	Progress   float64 `json:"progress"` // 0-100 overall progress
-	Speed      string  `json:"speed"`
-	FileIndex  int     `json:"file_index"`
-	TotalFiles int     `json:"total_files"`
-	Message    string  `json:"message"`
+	ModelID    string    `json:"model_id"`
+	NodeName   string    `json:"node_name"`
+	FileName   string    `json:"file_name"`
+	BytesSent  int64     `json:"bytes_sent"`
+	TotalBytes int64     `json:"total_bytes"`
+	Progress   float64   `json:"progress"` // 0-100 overall progress
+	Speed      string    `json:"speed"`
+	FileIndex  int       `json:"file_index"`
+	TotalFiles int       `json:"total_files"`
+	Message    string    `json:"message"`
 	StartedAt  time.Time `json:"started_at"`
 }
 
+const (
+	// stagingBroadcastInterval bounds how often byte-level UpdateFile ticks are
+	// re-broadcast to peers (leading-edge debounce). State transitions (Start,
+	// FileComplete, Complete) always publish so peers never miss them.
+	stagingBroadcastInterval = time.Second
+	// stagingRemoteTTL drops a mirrored (remote) op whose last update is older
+	// than this. NATS pub/sub is fire-and-forget, so a missed Done event would
+	// otherwise leave a phantom staging row on a peer forever; a live op
+	// refreshes its mirror at least every stagingBroadcastInterval.
+	stagingRemoteTTL = 60 * time.Second
+)
+
+// stagingEntry wraps a StagingStatus with the bookkeeping needed to keep peer
+// replicas consistent: whether this op is mirrored from a peer (remote) vs.
+// owned locally, when it was last updated (for remote-mirror expiry), and when
+// its byte progress was last broadcast (for debounce).
+type stagingEntry struct {
+	status    StagingStatus
+	remote    bool
+	updatedAt time.Time
+	lastPub   time.Time
+}
+
 // StagingTracker tracks active file staging operations in-memory.
 // Used by SmartRouter to publish progress and by /api/operations to surface it.
+//
+// In distributed mode each frontend replica runs its own tracker. The replica
+// performing a transfer owns the op locally and broadcasts progress over NATS
+// (SetPublisher); peers mirror it via ApplyRemote (SubscribeBroadcasts) so a
+// /api/operations poll that round-robins onto any replica surfaces the op.
 type StagingTracker struct {
-	mu     sync.RWMutex
-	active map[string]*StagingStatus
+	mu        sync.RWMutex
+	active    map[string]*stagingEntry
+	publisher messaging.Publisher
+}
+
+// StagingProgressEvent is the wire payload a frontend replica broadcasts on
+// SubjectStagingProgress so peer replicas can mirror a staging op they did not
+// originate. Done signals the op finished (peers drop their mirrored copy).
+type StagingProgressEvent struct {
+	ModelID string         `json:"model_id"`
+	Status  *StagingStatus `json:"status,omitempty"`
+	Done    bool           `json:"done"`
 }
 
 // NewStagingTracker creates a new tracker.
 func NewStagingTracker() *StagingTracker {
 	return &StagingTracker{
-		active: make(map[string]*StagingStatus),
+		active: make(map[string]*stagingEntry),
 	}
 }
 
+// SetPublisher wires the NATS publisher used to broadcast staging progress to
+// peer replicas. No-op publisher (nil) keeps the tracker standalone.
+func (t *StagingTracker) SetPublisher(p messaging.Publisher) {
+	t.mu.Lock()
+	defer t.mu.Unlock()
+	t.publisher = p
+}
+
+// SubscribeBroadcasts subscribes to peer replicas' staging-progress broadcasts
+// and mirrors them into this tracker, so /api/operations on any replica surfaces
+// staging ops it did not originate. Returns the subscription for cleanup.
+func (t *StagingTracker) SubscribeBroadcasts(nc messaging.MessagingClient) (messaging.Subscription, error) {
+	return messaging.SubscribeJSON(nc, messaging.SubjectStagingProgressWildcard, func(evt StagingProgressEvent) {
+		if evt.ModelID == "" {
+			return
+		}
+		t.ApplyRemote(evt)
+	})
+}
+
+// publishStaging emits an event to the per-model staging subject. The publisher
+// is captured by the caller under the lock and passed in, so publishing happens
+// outside the lock (a slow NATS link must not stall the staging copy loop).
+func publishStaging(p messaging.Publisher, evt StagingProgressEvent) {
+	if p == nil {
+		return
+	}
+	_ = p.Publish(messaging.SubjectStagingProgress(evt.ModelID), evt)
+}
+
 // Start registers a new staging operation for the given model.
 func (t *StagingTracker) Start(modelID, nodeName string, totalFiles int) {
 	t.mu.Lock()
-	defer t.mu.Unlock()
-	t.active[modelID] = &StagingStatus{
-		ModelID:    modelID,
-		NodeName:   nodeName,
-		TotalFiles: totalFiles,
-		StartedAt:  time.Now(),
-		Message:    "Preparing to stage model files",
+	e := &stagingEntry{
+		status: StagingStatus{
+			ModelID:    modelID,
+			NodeName:   nodeName,
+			TotalFiles: totalFiles,
+			StartedAt:  time.Now(),
+			Message:    "Preparing to stage model files",
+		},
+		updatedAt: time.Now(),
+		// lastPub stays zero so the first UpdateFile tick always broadcasts.
 	}
+	t.active[modelID] = e
+	pub := t.publisher
+	snap := e.status
+	t.mu.Unlock()
+
+	publishStaging(pub, StagingProgressEvent{ModelID: modelID, Status: &snap})
 }
 
 // UpdateFile updates the tracker with current file transfer progress.
 func (t *StagingTracker) UpdateFile(modelID, fileName string, fileIndex int, bytesSent, totalBytes int64, speed string) {
 	t.mu.Lock()
-	defer t.mu.Unlock()
-	s, ok := t.active[modelID]
+	e, ok := t.active[modelID]
 	if !ok {
+		t.mu.Unlock()
 		return
 	}
+	s := &e.status
 	s.FileName = fileName
 	s.FileIndex = fileIndex
 	s.BytesSent = bytesSent
@@ -79,52 +159,121 @@ func (t *StagingTracker) UpdateFile(modelID, fileName string, fileIndex int, byt
 	} else {
 		s.Message = fmt.Sprintf("Staging %s", fileName)
 	}
+
+	e.updatedAt = time.Now()
+	// Leading-edge debounce: byte ticks fire many times per second; only
+	// re-broadcast at most once per stagingBroadcastInterval.
+	var pub messaging.Publisher
+	var snap StagingStatus
+	if time.Since(e.lastPub) >= stagingBroadcastInterval {
+		e.lastPub = time.Now()
+		pub = t.publisher
+		snap = e.status
+	}
+	t.mu.Unlock()
+
+	if pub != nil {
+		publishStaging(pub, StagingProgressEvent{ModelID: modelID, Status: &snap})
+	}
 }
 
 // FileComplete marks a single file as done within a staging operation.
 func (t *StagingTracker) FileComplete(modelID string, fileIndex, totalFiles int) {
 	t.mu.Lock()
-	defer t.mu.Unlock()
-	s, ok := t.active[modelID]
+	e, ok := t.active[modelID]
 	if !ok {
+		t.mu.Unlock()
 		return
 	}
+	s := &e.status
 	if totalFiles > 0 {
 		s.Progress = float64(fileIndex) / float64(totalFiles) * 100
 	}
 	s.BytesSent = 0
 	s.TotalBytes = 0
 	s.Speed = ""
+	e.updatedAt = time.Now()
+	e.lastPub = time.Now()
+	pub := t.publisher
+	snap := e.status
+	t.mu.Unlock()
+
+	// Always broadcast a per-file completion so peers' progress bars advance.
+	publishStaging(pub, StagingProgressEvent{ModelID: modelID, Status: &snap})
 }
 
 // Complete removes a staging operation (it's done).
 func (t *StagingTracker) Complete(modelID string) {
 	t.mu.Lock()
-	defer t.mu.Unlock()
+	_, ok := t.active[modelID]
 	delete(t.active, modelID)
+	pub := t.publisher
+	t.mu.Unlock()
+
+	if ok {
+		// Tell peers to drop their mirrored copy.
+		publishStaging(pub, StagingProgressEvent{ModelID: modelID, Done: true})
+	}
 }
 
-// GetAll returns a snapshot of all active staging operations.
+// ApplyRemote merges a peer replica's staging broadcast into this tracker. It
+// never re-broadcasts (no echo loop). A locally-owned op is authoritative: a
+// remote event for the same model is ignored, so the origin replica receiving
+// its own broadcast (and any stray peer event) cannot clobber or delete it.
+func (t *StagingTracker) ApplyRemote(evt StagingProgressEvent) {
+	t.mu.Lock()
+	defer t.mu.Unlock()
+
+	if existing, ok := t.active[evt.ModelID]; ok && !existing.remote {
+		// We own this op locally — ignore peer chatter about it.
+		return
+	}
+	if evt.Done {
+		delete(t.active, evt.ModelID)
+		return
+	}
+	if evt.Status == nil {
+		return
+	}
+	t.active[evt.ModelID] = &stagingEntry{
+		status:    *evt.Status,
+		remote:    true,
+		updatedAt: time.Now(),
+	}
+}
+
+// GetAll returns a snapshot of all active staging operations. Stale remote
+// mirrors (a peer op whose Done event was missed) are pruned here so they don't
+// linger in the UI.
 func (t *StagingTracker) GetAll() map[string]StagingStatus {
-	t.mu.RLock()
-	defer t.mu.RUnlock()
+	t.mu.Lock()
+	defer t.mu.Unlock()
+	now := time.Now()
 	result := make(map[string]StagingStatus, len(t.active))
-	for k, v := range t.active {
-		result[k] = *v
+	for k, e := range t.active {
+		if e.remote && now.Sub(e.updatedAt) > stagingRemoteTTL {
+			delete(t.active, k)
+			continue
+		}
+		result[k] = e.status
 	}
 	return result
 }
 
-// Get returns the status of a specific staging operation, or nil if not active.
+// Get returns the status of a specific staging operation, or nil if not active
+// (or a stale remote mirror).
 func (t *StagingTracker) Get(modelID string) *StagingStatus {
 	t.mu.RLock()
 	defer t.mu.RUnlock()
-	s, ok := t.active[modelID]
+	e, ok := t.active[modelID]
 	if !ok {
 		return nil
 	}
-	copy := *s
-	return &copy
+	if e.remote && time.Since(e.updatedAt) > stagingRemoteTTL {
+		return nil
+	}
+	s := e.status
+	return &s
 }
 
 // StagingProgressCallback is called by file stagers to report byte-level progress.

core/services/nodes/staging_progress_broadcast_test.go

@@ -0,0 +1,109 @@
+package nodes
+
+import (
+	"encoding/json"
+
+	. "github.com/onsi/ginkgo/v2"
+	. "github.com/onsi/gomega"
+
+	"github.com/mudler/LocalAI/core/services/messaging"
+)
+
+// decodeStagingEvents extracts every StagingProgressEvent the fake messaging
+// client captured, in publish order.
+func decodeStagingEvents(mc *fakeMessagingClient) []StagingProgressEvent {
+	mc.mu.Lock()
+	defer mc.mu.Unlock()
+	var out []StagingProgressEvent
+	for _, p := range mc.published {
+		var evt StagingProgressEvent
+		if err := json.Unmarshal(p.Data, &evt); err != nil {
+			continue
+		}
+		if evt.ModelID == "" {
+			continue
+		}
+		out = append(out, evt)
+	}
+	return out
+}
+
+var _ = Describe("StagingTracker cross-replica broadcast", func() {
+	Context("when a publisher is wired (distributed mode)", func() {
+		It("broadcasts staging progress so a peer replica surfaces an op it did not originate", func() {
+			mc := &fakeMessagingClient{}
+			origin := NewStagingTracker()
+			origin.SetPublisher(mc)
+
+			origin.Start("model-x", "worker-1", 1)
+			origin.UpdateFile("model-x", "weights.gguf", 1, 5<<30, 10<<30, "100 MiB/s")
+
+			events := decodeStagingEvents(mc)
+			Expect(events).ToNot(BeEmpty(), "writes must be broadcast over NATS")
+			Expect(mc.published[0].Subject).To(Equal(messaging.SubjectStagingProgress("model-x")))
+
+			// A peer replica that never ran the op merges the broadcast.
+			peer := NewStagingTracker()
+			for _, evt := range events {
+				peer.ApplyRemote(evt)
+			}
+
+			all := peer.GetAll()
+			Expect(all).To(HaveKey("model-x"))
+			Expect(all["model-x"].NodeName).To(Equal("worker-1"))
+			Expect(all["model-x"].FileName).To(Equal("weights.gguf"))
+			Expect(all["model-x"].TotalBytes).To(Equal(int64(10 << 30)))
+		})
+
+		It("removes the op from the peer when the origin completes it", func() {
+			mc := &fakeMessagingClient{}
+			origin := NewStagingTracker()
+			origin.SetPublisher(mc)
+
+			origin.Start("model-x", "worker-1", 1)
+			origin.Complete("model-x")
+
+			peer := NewStagingTracker()
+			for _, evt := range decodeStagingEvents(mc) {
+				peer.ApplyRemote(evt)
+			}
+			Expect(peer.GetAll()).ToNot(HaveKey("model-x"))
+		})
+
+		It("does not let a peer broadcast clobber an op this replica is itself running", func() {
+			local := NewStagingTracker()
+			local.Start("model-x", "worker-local", 2)
+			local.UpdateFile("model-x", "weights.gguf", 1, 9<<30, 10<<30, "")
+
+			// A stray/older remote event for the SAME modelID must not overwrite
+			// the authoritative local state, nor delete it.
+			local.ApplyRemote(StagingProgressEvent{
+				ModelID: "model-x",
+				Status:  &StagingStatus{ModelID: "model-x", NodeName: "worker-other", FileName: "stale.gguf"},
+			})
+			local.ApplyRemote(StagingProgressEvent{ModelID: "model-x", Done: true})
+
+			all := local.GetAll()
+			Expect(all).To(HaveKey("model-x"))
+			Expect(all["model-x"].NodeName).To(Equal("worker-local"))
+			Expect(all["model-x"].FileName).To(Equal("weights.gguf"))
+		})
+	})
+
+	Context("when no publisher is wired (standalone mode)", func() {
+		It("does not broadcast", func() {
+			mc := &fakeMessagingClient{}
+			t := NewStagingTracker()
+			t.Start("model-x", "worker-1", 1)
+			t.UpdateFile("model-x", "weights.gguf", 1, 1<<30, 10<<30, "")
+			Expect(mc.published).To(BeEmpty())
+		})
+	})
+})
+
+var _ = Describe("SubjectStagingProgress", func() {
+	It("namespaces by model id and matches the wildcard prefix", func() {
+		Expect(messaging.SubjectStagingProgress("model-x")).To(Equal("staging.model-x.progress"))
+		Expect(messaging.SubjectStagingProgressWildcard).To(Equal("staging.*.progress"))
+	})
+})