chore(openclaw): refresh runtime models and credentials

litellm-init-models.sh

@@ -171,33 +171,33 @@ add_copilot_model "copilot-gemini-3.1-pro"       "gemini-3.1-pro-preview"
 add_copilot_model "copilot-grok-code-fast"       "grok-code-fast-1"
 
 # Local models (llama.cpp — no API key, custom model_info)
-if ! echo "$EXISTING" | grep -qx "qwen2.5-14b-local"; then
+if ! echo "$EXISTING" | grep -qx "gemma-3-12b-local"; then
   status=$(curl -s -o /tmp/model_resp.json -w "%{http_code}" \
     -X POST "$LITELLM_URL/model/new" \
     -H "Authorization: Bearer $LITELLM_MASTER_KEY" \
     -H "Content-Type: application/json" \
     -d '{
-      "model_name": "qwen2.5-14b-local",
+      "model_name": "gemma-3-12b-local",
       "litellm_params": {
-        "model": "openai/Qwen2.5-14B-Instruct-Q4_K_M.gguf",
+        "model": "openai/gemma-3-12b-it-q4_0.gguf",
         "api_base": "http://192.168.153.113:18806/v1"
       },
       "model_info": {
         "mode": "chat",
         "max_tokens": 8192,
-        "max_input_tokens": 32768,
-        "description": "Qwen2.5-14B-Instruct running locally via llama.cpp. Supports tool calling. Private/offline use.",
+        "max_input_tokens": 131072,
+        "description": "Gemma 3 12B Instruct running locally via llama.cpp. Private/offline use.",
         "supports_function_calling": true,
         "supports_vision": false
       }
     }')
   if [ "$status" = "200" ] || [ "$status" = "201" ]; then
-    echo "  [ok]   qwen2.5-14b-local"
+    echo "  [ok]   gemma-3-12b-local"
   else
-    echo "  [fail] qwen2.5-14b-local — HTTP $status: $(cat /tmp/model_resp.json)"
+    echo "  [fail] gemma-3-12b-local — HTTP $status: $(cat /tmp/model_resp.json)"
   fi
 else
-  echo "  [skip] qwen2.5-14b-local"
+  echo "  [skip] gemma-3-12b-local"
 fi
 
 echo "Done."