go back to last working version, but using gpt-40-mini

gaia_graph.py

@@ -11,9 +11,6 @@ from langchain.agents import initialize_agent, AgentType
 from langchain_openai import ChatOpenAI
 from langgraph.graph import StateGraph, END
 
-import openai  # Import openai to catch RateLimitError
-from tenacity import retry, wait_exponential, stop_after_attempt, retry_if_exception_type
-
 # ─── Load Environment Variables ──────────────────────────────────────────────
 load_dotenv()
 OPENAI_API_KEY = os.getenv("OPENAI_API_KEY")
@@ -78,7 +75,7 @@ search_tool = Tool(
 # ─── Create LLM Agent ────────────────────────────────────────────────────────
 llm = ChatOpenAI(
     temperature=0.0,
-    model="gpt-4o",
+    model="gpt-4o-mini",
     openai_api_key=OPENAI_API_KEY
 )
 
@@ -105,18 +102,9 @@ class GaiaState(TypedDict):
     question: str
     answer: str
 
-# ─── Retry on Rate Limit Error ───────────────────────────────────────────────
-@retry(
-    wait=wait_exponential(multiplier=1, min=1, max=10),
-    stop=stop_after_attempt(5),
-    retry=retry_if_exception_type(openai.error.RateLimitError)
-)
-def run_agent_with_retry(question: str) -> str:
-    return agent_executor.run(question)
-
 # ─── Define Node Function ────────────────────────────────────────────────────
 def agent_node(state: GaiaState) -> GaiaState:
-    raw = run_agent_with_retry(state["question"])
+    raw = agent_executor.run(state["question"])
     return {"question": state["question"], "answer": clean_answer(raw)}
 
 # ─── Build LangGraph ─────────────────────────────────────────────────────────