Change provider to OpenAI.

app.py

@@ -6,7 +6,8 @@ import mimetypes
 import requests
 import pandas as pd
 from llama_index.core.llms import ChatMessage, TextBlock, ImageBlock, AudioBlock
-from llama_index.llms.google_genai import GoogleGenAI
+# from llama_index.llms.google_genai import GoogleGenAI
+from llama_index.llms.openai import OpenAI
 from llama_index.core.agent.workflow import ReActAgent, AgentOutput
 from llama_index.core.tools import FunctionTool
 from llama_index.tools.duckduckgo import DuckDuckGoSearchToolSpec
@@ -27,7 +28,7 @@ load_dotenv()
 # --- Constants ---
 DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
 SYSTEM_PROMPT = (Path(__file__).parent / 'system_prompt.txt').read_text()
-GOOGLE_API_KEY = os.environ['GOOGLE_API_KEY']
+# GOOGLE_API_KEY = os.environ['GOOGLE_API_KEY']
 
 # --- Basic Agent Definition ---
 class BasicAgent:
@@ -37,7 +38,8 @@ class BasicAgent:
         wikipedia_search_tool = FunctionTool.from_defaults(WikipediaToolSpec().search_data)
         self._tools = [search_tool, wikipedia_load_tool, wikipedia_search_tool]
 
-        self._llm = GoogleGenAI(api_key=GOOGLE_API_KEY, model="gemini-2.0-flash", max_tokens=1600)
+        #self._llm = GoogleGenAI(api_key=GOOGLE_API_KEY, model="gemini-2.0-flash", max_tokens=1600)
+        self._llm = OpenAI(model="gpt-4o-mini")
         self._agent = ReActAgent(tools=self._tools, llm=self._llm)
         # Modify the react prompt.
         self._agent.update_prompts({"react_header": SYSTEM_PROMPT})
@@ -72,23 +74,28 @@ def fetch_questions(api_url: str = DEFAULT_API_URL):
 
 
 def get_media_type(filename: str):
-    media_type = mimetypes.guess_type(filename)[0]
-    if media_type is not None:
-        return media_type.split('/')[0]
+    media_type_and_format = mimetypes.guess_type(filename)[0]
+    if media_type_and_format is not None:
+        media_type, media_format = media_type_and_format.split('/')
+        if media_type == "audio" and media_format == "mpeg":
+            media_format = "mp3"
+        return media_type, media_format
+    else:
+        return None, None
 
 
 def get_media_content(item):
     if item.get('file_name'):
         file_response = requests.get(f"{DEFAULT_API_URL}/files/{item.get('task_id')}")
         if file_response:
-            media_type = get_media_type(item.get('file_name'))
+            media_type, media_format = get_media_type(item.get('file_name'))
             if media_type == 'image':
                 return ImageBlock(image=file_response.content)
             elif media_type == 'text':
                 return TextBlock(text=file_response.content)
             # Audio currently not supported?
             elif media_type == 'audio':
-                return AudioBlock(audio=file_response.content)
+                return AudioBlock(audio=file_response.content, format=media_format)
 
 
 def create_question_message(item):

requirements.txt

@@ -3,6 +3,6 @@ gradio
 requests
 pandas
 llama-index
-llama-index-llms-google-genai
+llama-index-llms-openai
 llama-index-tools-duckduckgo
 llama-index-tools-wikipedia