fix(gallery): correct Qwen3.8 MTP projector path

Use the Q8_0 projector already downloaded by the Q4 MTP entry.
The previous BF16 path pointed at a file absent from the entry.

Assisted-by: Codex:GPT-6

docs/content/features/model-gallery.md

@@ -43,6 +43,17 @@ Both views use the same model selection and 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.
 
+## Qwen3.8-27B with MTP
+
+Install the Q4_K_M build with MTP speculative decoding explicitly:
+
+```bash
+local-ai models install qwen3.8-27b-q4 --variant qwen3.8-27b-q4-mtp
+```
+
+This llama.cpp variant includes the Q4_0 draft model and Q8_0 vision projector
+for image input. Use a current llama.cpp backend with Qwen3.8 MTP support.
+
 ## Spark-X2.5-1.7B
 
 Install Spark-X2.5-1.7B with automatic selection between its Q4_K_M and Q8_0

gallery/index.yaml

@@ -4317,7 +4317,7 @@
     known_usecases:
       - chat
       - vision
-    mmproj: llama-cpp/mmproj/qwen3.8-27b/mmproj-Qwen3.8-27B-BF16.gguf
+    mmproj: llama-cpp/mmproj/qwen3.8-27b/mmproj-Qwen3.8-27B-Q8_0.gguf
     options:
       - use_jinja:true
       - spec_type:draft-mtp