Model: Locked clinical engine to Gemini 2.5 Flash-Lite (Free Tier Optimized)

backend/main.py

@@ -66,8 +66,8 @@ def _gemini(prompt: str, retries: int = 4, override_key: str = None) -> Optional
         "generationConfig": {"temperature": 0.1, "maxOutputTokens": 2048}
     }
     
-    # Try multiple models for compatibility (Gemini 3.1 Flash-Lite is latest/best for speed)
-    models = ["gemini-3.1-flash-lite", "gemini-2.0-flash", "gemini-1.5-flash"]
+    # Use Gemini 2.5 Flash-Lite as requested (Best for free tier)
+    models = ["gemini-2.5-flash-lite"]
     
     for model in models:
         last_error = ""