fix using gemini

agent.py

@@ -3,15 +3,23 @@ from smolagents import (
     ToolCallingAgent,
     InferenceClientModel,
     WebSearchTool,
+    OpenAIServerModel,
 )
 from prompts import prompt_templates
+import os
 
-model = InferenceClientModel(
-            max_tokens=2000,
-            temperature=0,
-            model_id='gemini-2.0-flash',
-            custom_role_conversions=None,
-        )
+#model = InferenceClientModel(
+#            max_tokens=2096,
+#            temperature=0.5,
+#            model_id='Qwen/Qwen2.5-Coder-32B-Instruct',
+#            custom_role_conversions=None,
+#        )
+
+model = OpenAIServerModel(
+    model_id="gemini-2.0-flash",
+    api_base="https://generativelanguage.googleapis.com/v1beta/openai/",
+    api_key=os.getenv("GEMINI_API_KEY"),
+)
 
 web_agent = ToolCallingAgent(
     tools=[WebSearchTool()],

requirements.txt

@@ -1,3 +1,3 @@
 gradio
 requests
-smolagents
+smolagents[openai]