feat(gallery): add LFM Turbo Brilliance variants

core/http/endpoints/mcp/localai_assistant_test.go

@@ -83,6 +83,10 @@
 	return "stub-backend-job", nil
 }
 
+func (stubClient) InstallLibraryBackend(_ context.Context, _ localaitools.InstallLibraryBackendRequest) (*localaitools.InstallLibraryBackendResult, error) {
+	return &localaitools.InstallLibraryBackendResult{Status: "installed", Name: "stub-backend", SourceID: "stub-source"}, nil
+}
+
 func (stubClient) UpgradeBackend(_ context.Context, _ string) (string, error) {
 	return "stub-upgrade-job", nil
 }

docs/content/features/model-gallery.md

@@ -1112,3 +1112,27 @@ is still running is reported as queued until the installer picks it up:
 A job ID is queryable from the moment `/models/apply` returns it, so a `404`/`500`
 from this endpoint means the ID is genuinely unknown rather than merely waiting
 its turn.
+
+## LFM2.5 Turbo Brilliance
+
+The gallery includes DavidAU's LFM2.5-24B-A2B Turbo Brilliance build for
+text chat and reasoning through llama.cpp. The IQ4_XS entry links to a
+Q8_0 MAX variant. Both use the embedded Turbo Brilliance chat template.
+
+Install with automatic variant selection:
+
+```bash
+local-ai models install lfm2.5-24b-a2b-turbo-brilliance-iq4-xs
+```
+
+Select Q8_0 MAX explicitly:
+
+```bash
+local-ai models install lfm2.5-24b-a2b-turbo-brilliance-iq4-xs --variant lfm2.5-24b-a2b-turbo-brilliance-q8
+```
+
+Both entries default to a 32,768-token context, temperature 0.1, top-k 50,
+and repetition penalty 1.05. The weights support up to 131,072 tokens;
+increase `context_size` in the installed configuration if needed.
+See the [publisher's model card](https://huggingface.co/DavidAU/LFM2.5-24B-A2B-Turbo-Brilliance-50X-NEO-MAX-Imatrix-GGUF)
+for the template's prompt controls and Liquid's LFM license.

gallery/index.yaml

@@ -3,6 +3,90 @@
 # written by scripts/models/cmd/modelcatalog and stays only so releases before
 # the cutover keep working. CI fails when it differs from the generated copy:
 #   go run ./scripts/models/cmd/modelcatalog --library gallery/models/index.yaml --out gallery/index.yaml
+- url: github:mudler/LocalAI/gallery/virtual.yaml@master
+  name: lfm2.5-24b-a2b-turbo-brilliance-iq4-xs
+  description: |
+    DavidAU's Turbo Brilliance build of LFM2.5-24B-A2B, a text-only reasoning
+    model with 24B total parameters and 2.45B active parameters. This entry
+    uses the IQ4_XS GGUF quantization and the embedded Turbo Brilliance
+    chat template for multilingual chat and tool use.
+  license: other
+  urls:
+    - https://huggingface.co/LiquidAI/LFM2.5-24B-A2B
+    - https://huggingface.co/DavidAU/LFM2.5-24B-A2B-Turbo-Brilliance-50X-NEO-MAX-Imatrix-GGUF
+  icon: https://cdn-uploads.huggingface.co/production/uploads/61b8e2ba285851687028d395/2b08LKpev0DNEk6DlnWkY.png
+  tags:
+    - llm
+    - gguf
+    - reasoning
+    - cpu
+    - gpu
+  files:
+    - filename: llama-cpp/models/lfm2.5-24b-a2b-turbo-brilliance/LFM2.5-24B-A2B-TBrill-NEO-IQ4_XS.gguf
+      sha256: a1562db04b263e5406ac8412e07be6ea92941ea653653e9f678ab13f430e59b1
+      uri: https://huggingface.co/DavidAU/LFM2.5-24B-A2B-Turbo-Brilliance-50X-NEO-MAX-Imatrix-GGUF/resolve/69107daec71f075c94d7d84929b4a0222d1b8446/LFM2.5-24B-A2B-TBrill-NEO-IQ4_XS.gguf
+  overrides:
+    backend: llama-cpp
+    context_size: 32768
+    function:
+      automatic_tool_parsing_fallback: true
+      grammar:
+        disable: true
+    known_usecases:
+      - chat
+      - completion
+    options:
+      - use_jinja:true
+    parameters:
+      model: llama-cpp/models/lfm2.5-24b-a2b-turbo-brilliance/LFM2.5-24B-A2B-TBrill-NEO-IQ4_XS.gguf
+      repeat_penalty: 1.05
+      temperature: 0.1
+      top_k: 50
+    template:
+      use_tokenizer_template: true
+  variants:
+    - model: lfm2.5-24b-a2b-turbo-brilliance-q8
+- url: github:mudler/LocalAI/gallery/virtual.yaml@master
+  name: lfm2.5-24b-a2b-turbo-brilliance-q8
+  description: |
+    DavidAU's Turbo Brilliance build of LFM2.5-24B-A2B, a text-only reasoning
+    model with 24B total parameters and 2.45B active parameters. This entry
+    uses the Q8_0 MAX GGUF quantization and the embedded Turbo Brilliance
+    chat template for multilingual chat and tool use.
+  license: other
+  urls:
+    - https://huggingface.co/LiquidAI/LFM2.5-24B-A2B
+    - https://huggingface.co/DavidAU/LFM2.5-24B-A2B-Turbo-Brilliance-50X-NEO-MAX-Imatrix-GGUF
+  icon: https://cdn-uploads.huggingface.co/production/uploads/61b8e2ba285851687028d395/2b08LKpev0DNEk6DlnWkY.png
+  tags:
+    - llm
+    - gguf
+    - reasoning
+    - cpu
+    - gpu
+  files:
+    - filename: llama-cpp/models/lfm2.5-24b-a2b-turbo-brilliance/LFM2.5-24B-A2B-TBrill-MAX-NEO-Q8_0.gguf
+      sha256: 17de10d567a0678dc6050445ecef3007ee1bf50df8d94a73d558d4133ad8aa4e
+      uri: https://huggingface.co/DavidAU/LFM2.5-24B-A2B-Turbo-Brilliance-50X-NEO-MAX-Imatrix-GGUF/resolve/69107daec71f075c94d7d84929b4a0222d1b8446/LFM2.5-24B-A2B-TBrill-MAX-NEO-Q8_0.gguf
+  overrides:
+    backend: llama-cpp
+    context_size: 32768
+    function:
+      automatic_tool_parsing_fallback: true
+      grammar:
+        disable: true
+    known_usecases:
+      - chat
+      - completion
+    options:
+      - use_jinja:true
+    parameters:
+      model: llama-cpp/models/lfm2.5-24b-a2b-turbo-brilliance/LFM2.5-24B-A2B-TBrill-MAX-NEO-Q8_0.gguf
+      repeat_penalty: 1.05
+      temperature: 0.1
+      top_k: 50
+    template:
+      use_tokenizer_template: true
 - url: github:mudler/LocalAI/gallery/virtual.yaml@master
   name: adavar-3b-q4
   description: |

gallery/models/index.yaml

@@ -4,6 +4,106 @@
 # changes and never edit it by hand. CI fails when it is stale. A build from
 # master reads it as the default model source; a tagged build reads the gallery
 # until the release cutover.
+- kind: model
+  id: lfm2.5-24b-a2b-turbo-brilliance-iq4-xs
+  name: lfm2.5-24b-a2b-turbo-brilliance-iq4-xs
+  version: 0.0.0-gallery.285918856e35
+  description: |
+    DavidAU's Turbo Brilliance build of LFM2.5-24B-A2B, a text-only reasoning
+    model with 24B total parameters and 2.45B active parameters. This entry
+    uses the IQ4_XS GGUF quantization and the embedded Turbo Brilliance
+    chat template for multilingual chat and tool use.
+  license: other
+  urls:
+    - https://huggingface.co/LiquidAI/LFM2.5-24B-A2B
+    - https://huggingface.co/DavidAU/LFM2.5-24B-A2B-Turbo-Brilliance-50X-NEO-MAX-Imatrix-GGUF
+  tags:
+    - llm
+    - gguf
+    - reasoning
+    - cpu
+    - gpu
+  icon: https://cdn-uploads.huggingface.co/production/uploads/61b8e2ba285851687028d395/2b08LKpev0DNEk6DlnWkY.png
+  channel: gallery
+  policy:
+    requireSignature: false
+  schemaVersion: 1
+  model:
+    url: github:mudler/LocalAI/gallery/virtual.yaml@master
+    files:
+      - filename: llama-cpp/models/lfm2.5-24b-a2b-turbo-brilliance/LFM2.5-24B-A2B-TBrill-NEO-IQ4_XS.gguf
+        sha256: a1562db04b263e5406ac8412e07be6ea92941ea653653e9f678ab13f430e59b1
+        uri: https://huggingface.co/DavidAU/LFM2.5-24B-A2B-Turbo-Brilliance-50X-NEO-MAX-Imatrix-GGUF/resolve/69107daec71f075c94d7d84929b4a0222d1b8446/LFM2.5-24B-A2B-TBrill-NEO-IQ4_XS.gguf
+    overrides:
+      backend: llama-cpp
+      context_size: 32768
+      function:
+        automatic_tool_parsing_fallback: true
+        grammar:
+          disable: true
+      known_usecases:
+        - chat
+        - completion
+      options:
+        - use_jinja:true
+      parameters:
+        model: llama-cpp/models/lfm2.5-24b-a2b-turbo-brilliance/LFM2.5-24B-A2B-TBrill-NEO-IQ4_XS.gguf
+        repeat_penalty: 1.05
+        temperature: 0.1
+        top_k: 50
+      template:
+        use_tokenizer_template: true
+    variants:
+      - model: lfm2.5-24b-a2b-turbo-brilliance-q8
+- kind: model
+  id: lfm2.5-24b-a2b-turbo-brilliance-q8
+  name: lfm2.5-24b-a2b-turbo-brilliance-q8
+  version: 0.0.0-gallery.401e9a9ccabf
+  description: |
+    DavidAU's Turbo Brilliance build of LFM2.5-24B-A2B, a text-only reasoning
+    model with 24B total parameters and 2.45B active parameters. This entry
+    uses the Q8_0 MAX GGUF quantization and the embedded Turbo Brilliance
+    chat template for multilingual chat and tool use.
+  license: other
+  urls:
+    - https://huggingface.co/LiquidAI/LFM2.5-24B-A2B
+    - https://huggingface.co/DavidAU/LFM2.5-24B-A2B-Turbo-Brilliance-50X-NEO-MAX-Imatrix-GGUF
+  tags:
+    - llm
+    - gguf
+    - reasoning
+    - cpu
+    - gpu
+  icon: https://cdn-uploads.huggingface.co/production/uploads/61b8e2ba285851687028d395/2b08LKpev0DNEk6DlnWkY.png
+  channel: gallery
+  policy:
+    requireSignature: false
+  schemaVersion: 1
+  model:
+    url: github:mudler/LocalAI/gallery/virtual.yaml@master
+    files:
+      - filename: llama-cpp/models/lfm2.5-24b-a2b-turbo-brilliance/LFM2.5-24B-A2B-TBrill-MAX-NEO-Q8_0.gguf
+        sha256: 17de10d567a0678dc6050445ecef3007ee1bf50df8d94a73d558d4133ad8aa4e
+        uri: https://huggingface.co/DavidAU/LFM2.5-24B-A2B-Turbo-Brilliance-50X-NEO-MAX-Imatrix-GGUF/resolve/69107daec71f075c94d7d84929b4a0222d1b8446/LFM2.5-24B-A2B-TBrill-MAX-NEO-Q8_0.gguf
+    overrides:
+      backend: llama-cpp
+      context_size: 32768
+      function:
+        automatic_tool_parsing_fallback: true
+        grammar:
+          disable: true
+      known_usecases:
+        - chat
+        - completion
+      options:
+        - use_jinja:true
+      parameters:
+        model: llama-cpp/models/lfm2.5-24b-a2b-turbo-brilliance/LFM2.5-24B-A2B-TBrill-MAX-NEO-Q8_0.gguf
+        repeat_penalty: 1.05
+        temperature: 0.1
+        top_k: 50
+      template:
+        use_tokenizer_template: true
 - kind: model
   id: adavar-3b-q4
   name: adavar-3b-q4