We need to send multiple prompts in one request for the chat endpoint

We need to have multiple prompts in one request for the chat endpoint.

This is so us programmers can use the chat endpoint in a similar fashion to the completions endpoint where we can currently send 20 prompts per request.

This would be very useful as sending a whole request per prompt is very inefficient.

Hi @joshuasy10

Is the completion for required for conversation or a singular prompt over the chat completion endpoint?

If I am understanding you correctly, I don’t think any context is required, however this is besides the point, I have a brief setup of an interaction, but I’d like to send these extra user prompts in bulk because it’s very inefficient using a hole request for each prompt. This is not supported with the https://api.openai.com/v1/chat/completions endpoint.


I have a suspicion this is what I’m supposed to use a ‘fine tuning’ model for but I haven’t properly looked into what that’s all about.

Edit: Had a look, fine-tuning isn’t yet available foe the gpt-3.5-turbo model so I guess that’s another feature request.