Update agent.py

agent.py

@@ -150,7 +150,9 @@ class GaiaAgent:
         
         # Rate limiting
         self.last_call_time = 0
-        self.min_delay = 1  # Groq is very fast, minimal delay needed
+        self.min_delay = 10  # 10 seconds between tasks to avoid rate limits
+        self.tokens_used_in_window = 0
+        self.window_start_time = time.time()
         
         # Initialize agent with primary model
         self._reinitialize_agent()
@@ -196,7 +198,7 @@ class GaiaAgent:
         print(f"🔹 Task ID: {task_id}")
         print(f"🔹 Question: {question[:100]}...")
         
-        # Try to get answer with retry logic
+        # Try to get answer with retry logic and exponential backoff
         answer = None
         for attempt in range(self.max_retries + 1):
             try:
@@ -204,13 +206,33 @@ class GaiaAgent:
                 if answer:
                     break
             except Exception as e:
-                print(f"⚠️ Attempt {attempt + 1} failed: {str(e)[:100]}")
-                if attempt < self.max_retries:
-                    print(f"🔄 Retrying with next model...")
-                    self._reinitialize_agent()
-                    time.sleep(2)
+                error_str = str(e)
+                print(f"⚠️ Attempt {attempt + 1} failed: {error_str[:150]}")
+                
+                # Check if it's a rate limit error
+                if "rate_limit" in error_str.lower() or "Rate limit" in error_str:
+                    # Extract wait time if available
+                    import re
+                    wait_match = re.search(r'(\d+\.?\d*)\s*s', error_str)
+                    if wait_match:
+                        wait_time = float(wait_match.group(1)) + 2  # Add 2s buffer
+                    else:
+                        wait_time = 15 * (attempt + 1)  # Exponential backoff: 15s, 30s, 45s
+                    
+                    print(f"⏳ Rate limit hit. Waiting {wait_time:.1f}s before retry...")
+                    time.sleep(wait_time)
+                    
+                    if attempt < self.max_retries:
+                        print(f"🔄 Retrying (attempt {attempt + 2}/{self.max_retries + 1})...")
+                        continue
                 else:
-                    answer = f"⚠️ Agent failed after {self.max_retries + 1} attempts: {e}"
+                    # Non-rate-limit error
+                    if attempt < self.max_retries:
+                        print(f"🔄 Retrying with fresh agent...")
+                        self._reinitialize_agent()
+                        time.sleep(2)
+                    else:
+                        answer = f"⚠️ Agent failed after {self.max_retries + 1} attempts: {e}"
         
         if not answer:
             answer = "⚠️ Sorry, I could not generate a valid response."