As I know it's impossible for current large language models to return "real-time" data, because you need to retrain it after every data update. And this will take time.