feat(gallery): add Jumbo Fusion 27B variants

Add pinned Q4_K_M and Q8_0 builds with linked variants and install docs.
Regenerate the gallery from the model library and fix its stale header.

Assisted-by: Codex:gpt-6

docs/content/features/model-gallery.md

@@ -44,6 +44,24 @@ Both views store the view, search, filter, and selection in the URL. Installing
 from Explore does not move you away from the catalog; the entry updates in place
 when the operation finishes.
 
+### Qwen3.8 Jumbo Fusion 27B
+
+The gallery includes DavidAU's Qwen3.8 Jumbo Fusion Q4_K_M and Q8_0 GGUF builds
+for text chat through llama.cpp. Both use the embedded chat template and a
+32,768-token default context. These entries do not include a vision projector
+or enable MTP decoding.
+
+Install with automatic variant selection:
+
+```bash
+local-ai models install qwen3.8-27b-jumbo-fusion-q4
+```
+
+Use `--variant qwen3.8-27b-jumbo-fusion-q4` or
+`--variant qwen3.8-27b-jumbo-fusion-q8` to select a specific build.
+See the [publisher's model card](https://huggingface.co/DavidAU/Qwen3.8-27B-Jumbo-Fusion-IQ-747-900-Turbo-Brilliance-50X-Reasoning-Instruct-NEO-MAX-MTP-GGUF)
+for its custom reasoning and instruction modes.
+
 ### MiniCPM-V 4.7 1B
 
 [MiniCPM-V 4.7 1B](https://huggingface.co/openbmb/MiniCPM-V-4.7-1B)

gallery/index.yaml

@@ -1087,6 +1087,66 @@
       model: Hemmingway-1.Q8_0.gguf
     template:
       use_tokenizer_template: true
+- url: github:mudler/LocalAI/gallery/qwen3.yaml@master
+  name: qwen3.8-27b-jumbo-fusion-q4
+  description: |
+    DavidAU's Jumbo Fusion edition of Qwen3.8-27B merges community fine-tunes
+    and includes a custom template with reasoning and instruction modes.
+    This Q4_K_M GGUF build supports text chat with a 32,768-token default context.
+  license: apache-2.0
+  urls:
+    - https://huggingface.co/Qwen/Qwen3.8-27B
+    - https://huggingface.co/DavidAU/Qwen3.8-27B-Jumbo-Fusion-IQ-747-900-Turbo-Brilliance-50X-Reasoning-Instruct-NEO-MAX-MTP-GGUF
+  tags:
+    - llm
+    - gguf
+    - cpu
+    - gpu
+    - reasoning
+  files:
+    - filename: Qwen3.8-27B-Jumbo-FIQ-747-900-T-Brill-50X-NEO-MTP-Q4_K_M.gguf
+      sha256: 10110a2cb45cf827f8deb6c17946376691697080e4ea0e3ae1922a7dd06fc713
+      uri: https://huggingface.co/DavidAU/Qwen3.8-27B-Jumbo-Fusion-IQ-747-900-Turbo-Brilliance-50X-Reasoning-Instruct-NEO-MAX-MTP-GGUF/resolve/63559b2ec40db5d17c383afb7e8574e038a1dc82/Qwen3.8-27B-Jumbo-FIQ-747-900-T-Brill-50X-NEO-MTP-Q4_K_M.gguf
+  overrides:
+    context_size: 32768
+    parameters:
+      min_p: 0.05
+      model: Qwen3.8-27B-Jumbo-FIQ-747-900-T-Brill-50X-NEO-MTP-Q4_K_M.gguf
+      repeat_penalty: 1
+      temperature: 0.7
+      top_k: 64
+      top_p: 0.95
+  variants:
+    - model: qwen3.8-27b-jumbo-fusion-q8
+- url: github:mudler/LocalAI/gallery/qwen3.yaml@master
+  name: qwen3.8-27b-jumbo-fusion-q8
+  description: |
+    DavidAU's Jumbo Fusion edition of Qwen3.8-27B merges community fine-tunes
+    and includes a custom template with reasoning and instruction modes.
+    This Q8_0 GGUF build supports text chat with a 32,768-token default context.
+  license: apache-2.0
+  urls:
+    - https://huggingface.co/Qwen/Qwen3.8-27B
+    - https://huggingface.co/DavidAU/Qwen3.8-27B-Jumbo-Fusion-IQ-747-900-Turbo-Brilliance-50X-Reasoning-Instruct-NEO-MAX-MTP-GGUF
+  tags:
+    - llm
+    - gguf
+    - cpu
+    - gpu
+    - reasoning
+  files:
+    - filename: Qwen3.8-27B-Jumbo-FIQ-747-900-T-Brill-50X-NEO-MTP-Q8_0.gguf
+      sha256: b18ae37279ebd34b92d2e15371cfe0a3aac490ccfc3f060ad9691c0aebfad6c1
+      uri: https://huggingface.co/DavidAU/Qwen3.8-27B-Jumbo-Fusion-IQ-747-900-Turbo-Brilliance-50X-Reasoning-Instruct-NEO-MAX-MTP-GGUF/resolve/63559b2ec40db5d17c383afb7e8574e038a1dc82/Qwen3.8-27B-Jumbo-FIQ-747-900-T-Brill-50X-NEO-MTP-Q8_0.gguf
+  overrides:
+    context_size: 32768
+    parameters:
+      min_p: 0.05
+      model: Qwen3.8-27B-Jumbo-FIQ-747-900-T-Brill-50X-NEO-MTP-Q8_0.gguf
+      repeat_penalty: 1
+      temperature: 0.7
+      top_k: 64
+      top_p: 0.95
 - url: github:mudler/LocalAI/gallery/qwen3.yaml@master
   name: qwen3.8-27b-turbo-brilliance-50x-q4
   description: |

gallery/models/index.yaml

@@ -1,9 +1,6 @@
-# Model library: the models of the model gallery in the library entry shape. The
-# gallery, gallery/index.yaml, stays the source of truth. This file is rendered
-# with scripts/models/cmd/modelcatalog; render it again after the gallery
-# 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.
+# Model library: the source of truth for models in the gallery.
+# Edit this file, then generate gallery/index.yaml with:
+#   go run ./scripts/models/cmd/modelcatalog --library gallery/models/index.yaml --out gallery/index.yaml
 - kind: model
   id: adavar-3b-q4
   name: adavar-3b-q4
@@ -1288,6 +1285,82 @@
         model: Hemmingway-1.Q8_0.gguf
       template:
         use_tokenizer_template: true
+- kind: model
+  id: qwen3.8-27b-jumbo-fusion-q4
+  name: qwen3.8-27b-jumbo-fusion-q4
+  version: 1.0.0
+  description: |
+    DavidAU's Jumbo Fusion edition of Qwen3.8-27B merges community fine-tunes
+    and includes a custom template with reasoning and instruction modes.
+    This Q4_K_M GGUF build supports text chat with a 32,768-token default context.
+  license: apache-2.0
+  urls:
+    - https://huggingface.co/Qwen/Qwen3.8-27B
+    - https://huggingface.co/DavidAU/Qwen3.8-27B-Jumbo-Fusion-IQ-747-900-Turbo-Brilliance-50X-Reasoning-Instruct-NEO-MAX-MTP-GGUF
+  tags:
+    - llm
+    - gguf
+    - cpu
+    - gpu
+    - reasoning
+  channel: gallery
+  policy:
+    requireSignature: false
+  schemaVersion: 1
+  model:
+    url: github:mudler/LocalAI/gallery/qwen3.yaml@master
+    files:
+      - filename: Qwen3.8-27B-Jumbo-FIQ-747-900-T-Brill-50X-NEO-MTP-Q4_K_M.gguf
+        sha256: 10110a2cb45cf827f8deb6c17946376691697080e4ea0e3ae1922a7dd06fc713
+        uri: https://huggingface.co/DavidAU/Qwen3.8-27B-Jumbo-Fusion-IQ-747-900-Turbo-Brilliance-50X-Reasoning-Instruct-NEO-MAX-MTP-GGUF/resolve/63559b2ec40db5d17c383afb7e8574e038a1dc82/Qwen3.8-27B-Jumbo-FIQ-747-900-T-Brill-50X-NEO-MTP-Q4_K_M.gguf
+    overrides:
+      context_size: 32768
+      parameters:
+        min_p: 0.05
+        model: Qwen3.8-27B-Jumbo-FIQ-747-900-T-Brill-50X-NEO-MTP-Q4_K_M.gguf
+        repeat_penalty: 1
+        temperature: 0.7
+        top_k: 64
+        top_p: 0.95
+    variants:
+      - model: qwen3.8-27b-jumbo-fusion-q8
+- kind: model
+  id: qwen3.8-27b-jumbo-fusion-q8
+  name: qwen3.8-27b-jumbo-fusion-q8
+  version: 1.0.0
+  description: |
+    DavidAU's Jumbo Fusion edition of Qwen3.8-27B merges community fine-tunes
+    and includes a custom template with reasoning and instruction modes.
+    This Q8_0 GGUF build supports text chat with a 32,768-token default context.
+  license: apache-2.0
+  urls:
+    - https://huggingface.co/Qwen/Qwen3.8-27B
+    - https://huggingface.co/DavidAU/Qwen3.8-27B-Jumbo-Fusion-IQ-747-900-Turbo-Brilliance-50X-Reasoning-Instruct-NEO-MAX-MTP-GGUF
+  tags:
+    - llm
+    - gguf
+    - cpu
+    - gpu
+    - reasoning
+  channel: gallery
+  policy:
+    requireSignature: false
+  schemaVersion: 1
+  model:
+    url: github:mudler/LocalAI/gallery/qwen3.yaml@master
+    files:
+      - filename: Qwen3.8-27B-Jumbo-FIQ-747-900-T-Brill-50X-NEO-MTP-Q8_0.gguf
+        sha256: b18ae37279ebd34b92d2e15371cfe0a3aac490ccfc3f060ad9691c0aebfad6c1
+        uri: https://huggingface.co/DavidAU/Qwen3.8-27B-Jumbo-Fusion-IQ-747-900-Turbo-Brilliance-50X-Reasoning-Instruct-NEO-MAX-MTP-GGUF/resolve/63559b2ec40db5d17c383afb7e8574e038a1dc82/Qwen3.8-27B-Jumbo-FIQ-747-900-T-Brill-50X-NEO-MTP-Q8_0.gguf
+    overrides:
+      context_size: 32768
+      parameters:
+        min_p: 0.05
+        model: Qwen3.8-27B-Jumbo-FIQ-747-900-T-Brill-50X-NEO-MTP-Q8_0.gguf
+        repeat_penalty: 1
+        temperature: 0.7
+        top_k: 64
+        top_p: 0.95
 - kind: model
   id: qwen3.8-27b-turbo-brilliance-50x-q4
   name: qwen3.8-27b-turbo-brilliance-50x-q4