Update app/core/model_loader.py

app/core/model_loader.py

@@ -95,8 +95,6 @@ class LLMSingleton:
             logger.info("📅 New day detected. Resetting daily AI quotas.")
             self.stats["date"] = today
             self.stats["daily_requests_count"] = 0
-            # We don't reset total_requests to keep lifetime stats, or we could. 
-            # Let's keep lifetime stats in 'total_requests' and daily in 'daily_requests_count'.
             self._save_stats()
 
     def _check_rpm_window(self):
@@ -141,7 +139,7 @@ class LLMSingleton:
 
         if not self.api_key:
             logger.error("Cannot generate: Missing GEMINI_API_KEY")
-            return ""
+            raise RuntimeError("MISSING_API_KEY")
 
         # Check daily quota before sending (Gemini 3 Flash Limit)
         if self.stats["daily_requests_count"] >= 500:
@@ -198,7 +196,7 @@ class LLMSingleton:
                     self.stats["errors"] += 1
                     self._save_stats()
                     logger.error(f"Gemini generation failed: {e}")
-                    return ""
+                    raise RuntimeError(f"GENERATION_FAILED: {e}")
         
         self.stats["errors"] += 1
         self._save_stats()