trying HuggingFace inference provider

app.py

@@ -1,6 +1,6 @@
-from smolagents import CodeAgent, load_tool, tool, InferenceClientModel
+from smolagents import CodeAgent, load_tool, tool, LiteLLMModel
 import datetime
-import requests
+import os
 import pytz
 import yaml
 from tools.final_answer import FinalAnswerTool
@@ -31,17 +31,12 @@ calories_checker = CaloriesCheckerTool()
 # If the agent does not answer, the model is overloaded, please use another model or the following Hugging Face Endpoint that also contains qwen2.5 coder:
 # model_id='https://pflgm2locj2t89co.us-east-1.aws.endpoints.huggingface.cloud'
 
-model = InferenceClientModel(
-    max_tokens=2048,
-    temperature=0.5,
-    # alternative model to test "Qwen/Qwen3-32B", "Qwen/Qwen2.5-Coder-7B-Instruct"
-    model_id="Qwen/Qwen2.5-Coder-7B-Instruct",
-    # model_id="Qwen/Qwen2.5-Coder-32B-Instruct",  # it is possible that this model may be overloaded
-    custom_role_conversions=None,
-    # inference_client_url="https://pflgm2locj2t89co.us-east-1.aws.endpoints.huggingface.cloud",
+gemini_api_key = os.environ.get("GEMINI_API_KEY", None)
+model = LiteLLMModel(
+    model_id="gemini/gemini-2.0-flash-lite",
+    api_key=gemini_api_key,
 )
 
-
 # Import tool from Hub
 image_generation_tool = load_tool("agents-course/text-to-image", trust_remote_code=True)
 

requirements.txt

@@ -5,3 +5,4 @@ duckduckgo_search==8.0.1
 pandas==2.2.3
 gradio==5.29.1
 gradio_client==1.10.1
+litellm==1.67.0.post1