Update app.py

app.py

@@ -48,7 +48,15 @@ async def chat_completions(request: Request):
                     yield f"data: {json.dumps(chunk)}\n\n"
                 yield "data: [DONE]\n\n"
 
-            return StreamingResponse(iter_content(), media_type="text/event-stream")
+            return StreamingResponse(
+                iter_content(),
+                media_type="text/event-stream",
+                headers={
+                    "Cache-Control": "no-cache",
+                    "Connection": "keep-alive",
+                    "X-Accel-Buffering": "no",
+                },
+            )
 
         return JSONResponse(content=output)
     except Exception as e: