mito-ai: fix default model

Author: aarondr77

mito-ai/src/components/ModelSelector.tsx

@@ -176,16 +176,24 @@ const ModelSelector: React.FC<ModelSelectorProps> = ({ onConfigChange }) => {
           }
         }
 
-        // Fallback to first available model if not found or invalid
+        // Fallback to default model or first available model if not found or invalid
         if (!fullModelName || !displayName || (fullModelName && !models.includes(fullModelName))) {
           if (models.length > 0) {
-            const firstModel = models[0];
-            fullModelName = firstModel;
-            // Check if it's a LiteLLM model
-            if (firstModel && firstModel.includes('/')) {
-              displayName = firstModel;
+            // First, try to use the default model if it's available
+            const defaultMapping = MODEL_MAPPINGS.find(m => m.displayName === DEFAULT_MODEL);
+            if (defaultMapping && models.includes(defaultMapping.fullName)) {
+              fullModelName = defaultMapping.fullName;
+              displayName = defaultMapping.displayName;
             } else {
-              displayName = MODEL_MAPPINGS.find(m => m.fullName === firstModel)?.displayName || firstModel;
+              // Fallback to first available model
+              const firstModel = models[0];
+              fullModelName = firstModel;
+              // Check if it's a LiteLLM model
+              if (firstModel && firstModel.includes('/')) {
+                displayName = firstModel;
+              } else {
+                displayName = MODEL_MAPPINGS.find(m => m.fullName === firstModel)?.displayName || firstModel;
+              }
             }
           } else {
             // Fallback to default if no models available

mito-ai/src/tests/AiChat/ModelSelector.test.tsx

@@ -95,4 +95,45 @@ describe('ModelSelector', () => {
       expect(mockOnConfigChange).toHaveBeenCalledWith(savedConfig);
     });
   });
+
+  it('defaults to default model when no storedConfig exists and GPT 4.1 is first in available models', async () => {
+    // Mock models with GPT 4.1 first (simulating the bug scenario)
+    mockRequestAPI.mockResolvedValue({
+      data: {
+        models: [
+          GPT_4_1_MODEL_NAME,
+          GPT_5_2_MODEL_NAME,
+          CLAUDE_SONNET_MODEL_NAME,
+          CLAUDE_HAIKU_MODEL_NAME,
+          GEMINI_3_FLASH_MODEL_NAME,
+          GEMINI_3_PRO_MODEL_NAME,
+        ]
+      }
+    });
+
+    // Ensure localStorage is empty (no storedConfig)
+    localStorage.clear();
+
+    render(<ModelSelector onConfigChange={mockOnConfigChange} />);
+
+    // Wait for models to load
+    await waitFor(() => {
+      expect(screen.queryByText('Loading models...')).not.toBeInTheDocument();
+    });
+
+    // Verify that the default model (Haiku 4.5) is selected, not GPT 4.1
+    expect(screen.getByText(DEFAULT_MODEL)).toBeInTheDocument();
+    
+    // Verify onConfigChange was called with Haiku model, not GPT 4.1
+    await waitFor(() => {
+      expect(mockOnConfigChange).toHaveBeenCalledWith({
+        model: DEFAULT_MODEL
+      });
+    });
+
+    // Ensure it was NOT called with GPT 4.1
+    expect(mockOnConfigChange).not.toHaveBeenCalledWith({
+      model: GPT_4_1_MODEL_NAME
+    });
+  });
 });