# I keep getting server_error from the chat completion API with gpt-3.5-turbo-16k model in the last few days. It does not happen all the time, but seems to happen in 80% of the cases. I use streaming and openai js library.