Say I have an Assistant chatbot and 5 users.
If all 5 users are interacting with the assistant isolated from each other (each has their own thread), would that impact the model in any way, especially with regards to its reliability and quality?
What if we scaled it to dozens of users, all at the same time (assuming rate limits were sufficient)?
An API assistant is just a set of instructions when you really look at it. They don’t really “exist”.
OpenAI has one “assistant” called ChatGPT servicing millions of users.
So the only impact would be your own code’s management, needing to make tons of network calls to a dozen methods for setup and use, and polling just to find out if a response is waiting.