Update main.py

main.py

@@ -184,6 +184,8 @@ class ChatRequest(BaseModel):
     model: str
     messages: List[Message]
 
+from fastapi.responses import Response
+
 @app.post("/v1/chat/completions")
 async def chat_completions(request: ChatRequest):
     messages = [{"role": msg.role, "content": msg.content} for msg in request.messages]
@@ -197,5 +199,5 @@ async def chat_completions(request: ChatRequest):
     async for chunk in async_generator:
         response_content += chunk if isinstance(chunk, str) else chunk.content
 
-    return {"content": response_content.strip()}  # Updated return
+    return Response(content=response_content.strip(), media_type="text/plain")  # Return plain text