Hello everyone, I’m currently developing a system that caters to multiple requests and I’m considering using several AI assistants. I’m wondering if there’s a maximum limit to the number of assistants I can create under my account. Can anyone shed some light on this?
An “assistant” is just a set of instructions really. You don’t need a different one to support hundreds or thousands of users. Just management of the messages, threads, and runs that are performed against them.
OpenAI hasn’t given any limit if you do want to make a bunch. The interface to manage them becomes unwieldy though, like if you want to list you can only get 100 at a time. A need to list a new starting point page after page that might change mid-process if this assistant creation is dynamic.
Attaching a file, the retrieval cost billed gets multiplied by every assistant on the same file (when its even working…)
The limit is the rate limits. Use of assistants can burn through tokens, and just a single run can be cut off and return an error on a low-tier account from unchecked internal iteration with no call rate mitigation or budget limitation like you’d program yourself with chat completions.