chore: bump llama.cpp to b5509 (#5120)

extensions/engine-management-extension/rolldown.config.mjs

@@ -15,7 +15,7 @@ export default defineConfig([
         `http://127.0.0.1:${process.env.CORTEX_API_PORT ?? '39291'}`
       ),
       PLATFORM: JSON.stringify(process.platform),
-      CORTEX_ENGINE_VERSION: JSON.stringify('b5488'),
+      CORTEX_ENGINE_VERSION: JSON.stringify('b5509'),
       DEFAULT_REMOTE_ENGINES: JSON.stringify(engines),
       DEFAULT_REMOTE_MODELS: JSON.stringify(models),
       DEFAULT_REQUEST_PAYLOAD_TRANSFORM: JSON.stringify(
@@ -38,7 +38,7 @@ export default defineConfig([
       file: 'dist/node/index.cjs.js',
     },
     define: {
-      CORTEX_ENGINE_VERSION: JSON.stringify('b5488'),
+      CORTEX_ENGINE_VERSION: JSON.stringify('b5509'),
     },
   },
 ])

extensions/inference-cortex-extension/download.bat

@@ -2,7 +2,7 @@
 set BIN_PATH=./bin
 set SHARED_PATH=./../../electron/shared
 set /p CORTEX_VERSION=<./bin/version.txt
-set ENGINE_VERSION=b5488
+set ENGINE_VERSION=b5509
 
 @REM Download llama.cpp binaries
 set DOWNLOAD_URL=https://github.com/menloresearch/llama.cpp/releases/download/%ENGINE_VERSION%/llama-%ENGINE_VERSION%-bin-win

extensions/inference-cortex-extension/download.sh

@@ -2,7 +2,7 @@
 
 # Read CORTEX_VERSION
 CORTEX_VERSION=$(cat ./bin/version.txt)
-ENGINE_VERSION=b5488
+ENGINE_VERSION=b5509
 CORTEX_RELEASE_URL="https://github.com/menloresearch/cortex.cpp/releases/download"
 ENGINE_DOWNLOAD_URL=https://github.com/menloresearch/llama.cpp/releases/download/${ENGINE_VERSION}/llama-${ENGINE_VERSION}-bin
 CUDA_DOWNLOAD_URL=https://github.com/menloresearch/llama.cpp/releases/download/${ENGINE_VERSION}

extensions/inference-cortex-extension/rolldown.config.mjs

@@ -19,7 +19,7 @@ export default defineConfig([
       CORTEX_SOCKET_URL: JSON.stringify(
         `ws://127.0.0.1:${process.env.CORTEX_API_PORT ?? '39291'}`
       ),
-      CORTEX_ENGINE_VERSION: JSON.stringify('b5488'),
+      CORTEX_ENGINE_VERSION: JSON.stringify('b5509'),
     },
   },
   {

src-tauri/binaries/download.bat

@@ -1,7 +1,7 @@
 @echo off
 
 set CORTEX_VERSION=1.0.13-rc6
-set ENGINE_VERSION=b5488
+set ENGINE_VERSION=b5509
 set ENGINE_DOWNLOAD_URL=https://github.com/menloresearch/llama.cpp/releases/download/%ENGINE_VERSION%/llama-%ENGINE_VERSION%-bin-win
 set ENGINE_DOWNLOAD_GGML_URL=https://github.com/ggml-org/llama.cpp/releases/download/%ENGINE_VERSION%/llama-%ENGINE_VERSION%-bin-win
 set CUDA_DOWNLOAD_URL=https://github.com/menloresearch/llama.cpp/releases/download/%ENGINE_VERSION%

src-tauri/binaries/download.sh

@@ -15,7 +15,7 @@ download() {
 
 # Read CORTEX_VERSION
 CORTEX_VERSION=1.0.13-rc6
-ENGINE_VERSION=b5488
+ENGINE_VERSION=b5509
 CORTEX_RELEASE_URL="https://github.com/menloresearch/cortex.cpp/releases/download"
 ENGINE_DOWNLOAD_URL=https://github.com/menloresearch/llama.cpp/releases/download/${ENGINE_VERSION}/llama-${ENGINE_VERSION}-bin
 CUDA_DOWNLOAD_URL=https://github.com/menloresearch/llama.cpp/releases/download/${ENGINE_VERSION}

web-app/src/containers/DropdownModelProvider.tsx

@@ -71,9 +71,11 @@ const DropdownModelProvider = ({ model }: DropdownModelProviderProps) => {
               title={displayModel}
               className="font-medium cursor-pointer flex items-center gap-1.5 relative z-20 max-w-38"
             >
+              {provider && (
                 <div className="shrink-0">
-                <ProvidersAvatar provider={provider as ProviderObject} />
+                  <ProvidersAvatar provider={provider} />
                 </div>
+              )}
               <span
                 className={cn(
                   'text-main-view-fg/80 truncate leading-normal',
@@ -84,11 +86,8 @@ const DropdownModelProvider = ({ model }: DropdownModelProviderProps) => {
               </span>
             </button>
           </DropdownMenuTrigger>
-          {currentModel?.settings && (
+          {currentModel?.settings && provider && (
-            <ModelSetting
+            <ModelSetting model={currentModel as Model} provider={provider} />
-              model={currentModel as Model}
-              provider={provider as ProviderObject}
-            />
           )}
         </div>
         <DropdownMenuContent

web-app/src/hooks/useChat.ts

@@ -25,6 +25,7 @@ import { getTools } from '@/services/mcp'
 import { MCPTool } from '@/types/completion'
 import { listen } from '@tauri-apps/api/event'
 import { SystemEvent } from '@/types/events'
+import { stopModel } from '@/services/models'
 
 export const useChat = () => {
   const { prompt, setPrompt } = usePrompt()
@@ -174,6 +175,17 @@ export const useChat = () => {
               }
             }
           }
+          // TODO: Remove this check when integrating new llama.cpp extension
+          if (
+            accumulatedText.length === 0 &&
+            toolCalls.length === 0 &&
+            activeThread.model?.id &&
+            provider.provider === 'llama.cpp'
+          ) {
+            await stopModel(activeThread.model.id, 'cortex')
+            throw new Error('No response received from the model')
+          }
+
           // Create a final content object for adding to the thread
           const finalContent = newAssistantThreadContent(
             activeThread.id,