Increase max_tokens to 700 and adjust temperature to 0.5 for educational tutor balance

backend/main.py

@@ -122,8 +122,8 @@ def run_flow_b(message: str, api_key: str, history: Optional[List[ChatMessage]]
     llm = ChatGoogleGenerativeAI(
         model="gemini-3.1-flash-lite",
         google_api_key=api_key,
-        temperature=0.4,
-        max_tokens=350,
+        temperature=0.5,
+        max_tokens=700,
         generation_config={"response_mime_type": "application/json"}
     )