try litellmmodel

app.py

@@ -5,7 +5,7 @@ import requests
 import inspect
 import pandas as pd
 from typing import Dict, List
-from smolagents import CodeAgent, tool, InferenceClientModel
+from smolagents import CodeAgent, LiteLLMModel, tool, InferenceClientModel, WikipediaSearchTool, VisitWebpageTool, DuckDuckGoSearchTool
 from langchain_community.tools.tavily_search import TavilySearchResults
 from langchain_community.document_loaders import WikipediaLoader, ArxivLoader
 
@@ -93,13 +93,49 @@ def wikipedia_search(query: str, load_max_docs: int=3) -> str:
 # ----- THIS IS WERE YOU CAN BUILD WHAT YOU WANT ------
 class BasicAgent:
     def __init__(self):
-        model_id = "Qwen/Qwen3-32B"
+        
+        model = LiteLLMModel(model_id=config.CLAUDE_3_5_SONNET_MODEL_ID, api_key=ANTHROPIC_API_KEY,
+                             temperature=DEFAULT_TEMPERATURE,
+                             max_retries=3)
+                             
+        # System prompt recommended by GAIA benchmark instruction https://huggingface.co/spaces/gaia-benchmark/leaderboard
+        # Emitting the part of final answer, as instructed by the HF course
+        system_prompt = """
+                         You are a general AI assistant.
+                         I will ask you a question.
+                         Report your thoughts.
+                         YOUR FINAL ANSWER should be a number OR as few words as possible OR a comma separated list of numbers and/or strings.
+                         If you are asked for a number, don't use comma to write your number neither use units such as $ or percent sign unless specified otherwise.
+                         If you are asked for a string, don't use articles, neither abbreviations (e.g. for cities), and write the digits in plain text unless specified otherwise.
+                         If you are asked for a comma separated list, apply the above rules depending of whether the element to be put in the list is a number or a string.
+                        """
+        
+        
+        # # Init agent with InferenceClientModel
+        # model_id = "Qwen/Qwen3-32B"
+        # self.agent = CodeAgent(
+        #     model=InferenceClientModel(model_id=model_id, token=os.getenv("HF_TOKEN")),
+        #     tools=[multiply, add, subtract, modulus, web_search, arxiv_search, wikipedia_search],
+        #     max_steps=10,
+        #     verbosity_level=2,
+        # )
+
+        # Init agent with LiteLLMModel
+        model = LiteLLMModel(model_id="openai/gpt-4o-mini", api_key=os.getenv("OPENAI_API_KEY"),
+                            #  temperature=DEFAULT_TEMPERATURE,
+                             max_retries=3)
         self.agent = CodeAgent(
-            model=InferenceClientModel(model_id=model_id, token=os.getenv("HF_TOKEN")),
-            tools=[multiply, add, subtract, modulus, web_search, arxiv_search, wikipedia_search],
-            max_steps=10,
-            verbosity_level=2,
+            tools=[
+                multiply, add, subtract, modulus, web_search, arxiv_search, wikipedia_search,
+                WikipediaSearchTool(),
+                DuckDuckGoSearchTool(),
+                VisitWebpageTool()
+            ],
+            model=model,
+            # planning_interval=DEFAULT_PLANNING_INTERVAL,
+            # step_callbacks=[generation_error_interrupter]
         )
+
         print("BasicAgent initialized.")
     def __call__(self, question: str) -> str:
         print(f"Agent received question (first 50 chars): {question[:50]}...")