feat(provider/llamacpp): update default larger local model to Qwen3.8-27B-Uncensored-Aggressive-Q4_K_M.gguf

config.sample.yml

@@ -19,8 +19,8 @@ llm:
   # Examples:
   # default: openai/gpt-5.6-sol
   # default: openrouter/z-ai/glm-5.2
-  # default: llamacpp/Qwen3.6-35B-A3B-Uncensored-HauhauCS-Aggressive-Q4_K_M.gguf
-  # default: sglang/Qwen3.6-35B-A3B-Uncensored-HauhauCS-Aggressive-Q4_K_M.gguf
+  # default: llamacpp/Qwen3.8-27B-Uncensored-Aggressive-Q4_K_M.gguf
+  # default: sglang/Qwen3.8-27B-Uncensored-Aggressive-Q4_K_M.gguf
   # default: /absolute/path/model.gguf
   default: null
   # Optional per-mode overrides. Null values fall back to llm.default.

scripts/setup/setup-local-llm.js

@@ -25,11 +25,11 @@ import { createSetupStatus } from './setup-status'
 const DEFAULT_LLM_OPTIONS = [
   {
     minimumTotalVRAM: LLM_HIGH_TIER_MINIMUM_TOTAL_VRAM,
-    name: 'Qwen3.6-35B-A3B-Uncensored-HauhauCS-Aggressive',
+    name: 'Qwen3.8-27B-Uncensored-Aggressive',
     version: 'Q4_K_M',
-    fileName: 'Qwen3.6-35B-A3B-Uncensored-HauhauCS-Aggressive-Q4_K_M.gguf',
+    fileName: 'Qwen3.8-27B-Uncensored-Aggressive-Q4_K_M.gguf',
     downloadURL:
-      'https://huggingface.co/HauhauCS/Qwen3.6-35B-A3B-Uncensored-HauhauCS-Aggressive/resolve/main/Qwen3.6-35B-A3B-Uncensored-HauhauCS-Aggressive-Q4_K_M.gguf?download=true'
+      'https://huggingface.co/0xKitkat/Qwen3.8-27B-Uncensored-Aggressive/resolve/main/Qwen3.8-27B-Uncensored-Aggressive-Q4_K_M.gguf?download=true'
   },
   {
     minimumTotalVRAM: LLM_MINIMUM_TOTAL_VRAM,

server/src/core/llm-manager/llm-providers/llamacpp-llm-provider.ts

@@ -794,6 +794,7 @@ export default class LlamaCPPLLMProvider extends AISDKRemoteLLMProvider {
         String(getURLPort(runtimeServerURL)),
         '--ctx-size',
         String(LOCAL_LLM_CONTEXT_WINDOW_TOKENS),
+        '--jinja',
         '--flash-attn',
         'on',
         '--cache-type-k',