I am planning to use threads to maintain the user’s previous context to answer in the best possible way based on my vector data that is already there, Also I want to know what will be the costing for this, how it will be calculated, and also how many levels, threads consider previous conversations
Have you read the pricing page?:
https://openai.com/api/pricing/
It’s hard to estimate because the assistant can burn tokens iterating until it decides it has reached a response it can share.
I think the best approach is to prototype your system and watch the dashboard then estimate how that would scale with more broad use.
1 Like