feat: materialize Hugging Face model artifacts (#10825)

* feat(config): add model artifact source contract Assisted-by: Codex:GPT-5 [Codex] * feat(downloader): add authenticated raw-byte progress Assisted-by: Codex:GPT-5 [Codex] * feat(huggingface): resolve immutable snapshot manifests Assisted-by: Codex:GPT-5 [Codex] * feat(models): add artifact storage primitives Assisted-by: Codex:GPT-5 [Codex] * feat(models): materialize pinned Hugging Face snapshots Assisted-by: Codex:GPT-5 [Codex] * feat(models): bind managed snapshots at runtime Assisted-by: Codex:GPT-5 [Codex] * feat(gallery): materialize model artifacts during install Assisted-by: Codex:GPT-5 [Codex] * feat(gallery): declare managed Hugging Face artifacts Assisted-by: Codex:GPT-5 [Codex] * feat(models): preload managed model artifacts Assisted-by: Codex:GPT-5 [Codex] * fix(gallery): retain shared artifact caches on delete Assisted-by: Codex:GPT-5 [Codex] * feat(models): report artifact acquisition progress Assisted-by: Codex:GPT-5 [Codex] * refactor(backends): load managed models from ModelFile Assisted-by: Codex:GPT-5 [Codex] * refactor(backends): load staged speech model snapshots Assisted-by: Codex:GPT-5 [Codex] * refactor(backends): use staged snapshots in engine backends Assisted-by: Codex:GPT-5 [Codex] * test(distributed): cover staged artifact snapshots Assisted-by: Codex:GPT-5 [Codex] * docs: explain managed model artifacts Assisted-by: Codex:GPT-5 [Codex] * docs: add product design context Assisted-by: Codex:GPT-5 [Codex] * feat(ui): show model artifact download progress Assisted-by: Codex:GPT-5 [Codex] * Eagerly materialize Hugging Face artifacts Materialize HF-backed model references as managed GGUF artifacts during load, with lazy download retained only as fallback. Assisted-by: Codex:GPT-5 [shell] * Refactor HF downloads through a shared executor Assisted-by: Codex:GPT-5 [shell] * drop Signed-off-by: Ettore Di Giacinto <mudler@localai.io> --------- Signed-off-by: Ettore Di Giacinto <mudler@localai.io> Co-authored-by: Ettore Di Giacinto <mudler@localai.io>
2026-07-30 09:57:57 -04:00 · 2026-07-15 01:09:33 +02:00
parent d82c38ee77
commit bcc41219f7
83 changed files with 4315 additions and 339 deletions
--- a/gallery/index.yaml
+++ b/gallery/index.yaml
@@ -2424,6 +2424,12 @@
      - torch_dtype:bf16
    parameters:
      model: lodestones/Chroma1-HD
+    artifacts:
+      - name: model
+        target: model
+        source:
+          type: huggingface
+          repo: lodestones/Chroma1-HD
    step: 40
 - name: nemotron-3-nano-omni-30b-a3b-reasoning-apex
  url: github:mudler/LocalAI/gallery/virtual.yaml@master
@@ -2473,6 +2479,12 @@
      temperature: 1
      top_k: -1
      top_p: 1
+    artifacts:
+      - name: model
+        target: model
+        source:
+          type: huggingface
+          repo: mudler/Nemotron-3-Nano-Omni-30B-A3B-Reasoning-APEX-GGUF
    template:
      use_tokenizer_template: true
 - name: carnice-v2-27b
@@ -3890,6 +3902,14 @@
      - transcript
    parameters:
      model: nvidia/parakeet-tdt-0.6b-v3
+    artifacts:
+      - name: model
+        target: model
+        source:
+          type: huggingface
+          repo: nvidia/parakeet-tdt-0.6b-v3
+          allow_patterns:
+            - "*.nemo"
 - name: voxtral-mini-4b-realtime
  url: github:mudler/LocalAI/gallery/virtual.yaml@master
  urls:
@@ -5080,6 +5100,12 @@
      voice_cloning: true
    parameters:
      model: openbmb/VoxCPM1.5
+    artifacts:
+      - name: model
+        target: model
+        source:
+          type: huggingface
+          repo: openbmb/VoxCPM1.5
 - name: neutts-air
  url: github:mudler/LocalAI/gallery/virtual.yaml@master
  urls:
@@ -5124,6 +5150,12 @@
      - image
    parameters:
      model: Tongyi-MAI/Z-Image-Turbo
+    artifacts:
+      - name: model
+        target: model
+        source:
+          type: huggingface
+          repo: Tongyi-MAI/Z-Image-Turbo
 - name: vllm-omni-wan2.2-t2v
  url: github:mudler/LocalAI/gallery/virtual.yaml@master
  urls:
@@ -5146,6 +5178,12 @@
      - video
    parameters:
      model: Wan-AI/Wan2.2-T2V-A14B-Diffusers
+    artifacts:
+      - name: model
+        target: model
+        source:
+          type: huggingface
+          repo: Wan-AI/Wan2.2-T2V-A14B-Diffusers
 - name: vllm-omni-wan2.2-i2v
  url: github:mudler/LocalAI/gallery/virtual.yaml@master
  urls:
@@ -5168,6 +5206,12 @@
      - video
    parameters:
      model: Wan-AI/Wan2.2-I2V-A14B-Diffusers
+    artifacts:
+      - name: model
+        target: model
+        source:
+          type: huggingface
+          repo: Wan-AI/Wan2.2-I2V-A14B-Diffusers
 - name: longcat-video
  url: github:mudler/LocalAI/gallery/virtual.yaml@master
  urls:
@@ -5273,6 +5317,12 @@
      - tts
    parameters:
      model: Qwen/Qwen3-Omni-30B-A3B-Instruct
+    artifacts:
+      - name: model
+        target: model
+        source:
+          type: huggingface
+          repo: Qwen/Qwen3-Omni-30B-A3B-Instruct
 - name: vllm-omni-qwen3-tts-custom-voice
  url: github:mudler/LocalAI/gallery/virtual.yaml@master
  urls:
@@ -5296,6 +5346,12 @@
      - tts
    parameters:
      model: Qwen/Qwen3-TTS-12Hz-1.7B-CustomVoice
+    artifacts:
+      - name: model
+        target: model
+        source:
+          type: huggingface
+          repo: Qwen/Qwen3-TTS-12Hz-1.7B-CustomVoice
 - name: ace-step-turbo
  url: github:mudler/LocalAI/gallery/virtual.yaml@master
  urls:
@@ -6027,6 +6083,12 @@
      - torch_dtype:bf16
    parameters:
      model: Tongyi-MAI/Z-Image
+    artifacts:
+      - name: model
+        target: model
+        source:
+          type: huggingface
+          repo: Tongyi-MAI/Z-Image
    step: 35
 - name: z-image-turbo-diffusers
  url: github:mudler/LocalAI/gallery/virtual.yaml@master
@@ -6052,6 +6114,12 @@
      - torch_dtype:bf16
    parameters:
      model: Tongyi-MAI/Z-Image-Turbo
+    artifacts:
+      - name: model
+        target: model
+        source:
+          type: huggingface
+          repo: Tongyi-MAI/Z-Image-Turbo
    step: 9
 - name: glm-4.7-flash-derestricted
  url: github:mudler/LocalAI/gallery/virtual.yaml@master
@@ -6110,6 +6178,12 @@
      - tts
    parameters:
      model: Qwen/Qwen3-TTS-12Hz-1.7B-CustomVoice
+    artifacts:
+      - name: model
+        target: model
+        source:
+          type: huggingface
+          repo: Qwen/Qwen3-TTS-12Hz-1.7B-CustomVoice
    tts:
      voice: Aiden
 - name: qwen3-tts-0.6b-custom-voice
@@ -6130,6 +6204,12 @@
      - tts
    parameters:
      model: Qwen/Qwen3-TTS-12Hz-0.6B-CustomVoice
+    artifacts:
+      - name: model
+        target: model
+        source:
+          type: huggingface
+          repo: Qwen/Qwen3-TTS-12Hz-0.6B-CustomVoice
    tts:
      voice: Aiden
 - name: fish-speech-s2-pro
@@ -6153,6 +6233,12 @@
      voice_cloning: true
    parameters:
      model: fishaudio/s2-pro
+    artifacts:
+      - name: model
+        target: model
+        source:
+          type: huggingface
+          repo: fishaudio/s2-pro
 - name: qwen3-asr-1.7b
  url: github:mudler/LocalAI/gallery/virtual.yaml@master
  urls:
@@ -6171,6 +6257,12 @@
      - transcript
    parameters:
      model: Qwen/Qwen3-ASR-1.7B
+    artifacts:
+      - name: model
+        target: model
+        source:
+          type: huggingface
+          repo: Qwen/Qwen3-ASR-1.7B
 - name: qwen3-asr-0.6b
  url: github:mudler/LocalAI/gallery/virtual.yaml@master
  urls:
@@ -6189,6 +6281,12 @@
      - transcript
    parameters:
      model: Qwen/Qwen3-ASR-0.6B
+    artifacts:
+      - name: model
+        target: model
+        source:
+          type: huggingface
+          repo: Qwen/Qwen3-ASR-0.6B
 - name: huihui-glm-4.7-flash-abliterated-i1
  url: github:mudler/LocalAI/gallery/virtual.yaml@master
  urls:
@@ -7635,6 +7733,12 @@
      - transcript
    parameters:
      model: Qwen/Qwen3-ASR-0.6B
+    artifacts:
+      - name: model
+        target: model
+        source:
+          type: huggingface
+          repo: Qwen/Qwen3-ASR-0.6B
 - name: qwen3-asr-1.7b
  url: github:mudler/LocalAI/gallery/virtual.yaml@master
  urls:
@@ -7653,6 +7757,12 @@
      - transcript
    parameters:
      model: Qwen/Qwen3-ASR-1.7B
+    artifacts:
+      - name: model
+        target: model
+        source:
+          type: huggingface
+          repo: Qwen/Qwen3-ASR-1.7B
 - name: glm-ocr
  url: github:mudler/LocalAI/gallery/virtual.yaml@master
  urls:
@@ -8830,6 +8940,12 @@
      - torch_dtype:bf16
    parameters:
      model: Lightricks/LTX-2
+    artifacts:
+      - name: model
+        target: model
+        source:
+          type: huggingface
+          repo: Lightricks/LTX-2
 - name: gpt-oss-20b
  url: github:mudler/LocalAI/gallery/harmony.yaml@master
  urls:
@@ -9078,6 +9194,12 @@
    name: dia
    parameters:
      model: nari-labs/Dia-1.6B-0626
+    artifacts:
+      - name: model
+        target: model
+        source:
+          type: huggingface
+          repo: nari-labs/Dia-1.6B-0626
    type: DiaForConditionalGeneration
 - name: outetts
  url: github:mudler/LocalAI/gallery/virtual.yaml@master
@@ -35853,6 +35975,12 @@
      - torch_dtype:bf16
    parameters:
      model: Lightricks/LTX-2.3
+    artifacts:
+      - name: model
+        target: model
+        source:
+          type: huggingface
+          repo: Lightricks/LTX-2.3
 - &ltx-2-3-dev-ggml
  name: ltx-2.3-22b-dev-ggml
  url: github:mudler/LocalAI/gallery/ltx-ggml.yaml@master
--- a/gallery/liquid-audio.yaml
+++ b/gallery/liquid-audio.yaml
@@ -31,6 +31,12 @@ config_file: |
    - vad
  parameters:
    model: LiquidAI/LFM2.5-Audio-1.5B
+  artifacts:
+    - name: model
+      target: model
+      source:
+        type: huggingface
+        repo: LiquidAI/LFM2.5-Audio-1.5B
  # Special tokens emitted in the text track during interleaved generation.
  # Included so a future client-side parser can spot them; the LFM2 tool-call
  # format itself is auto-detected by the upstream llama.cpp parser when the