Working with Batch API while keeping sessions


I’ve created an application using GPT API where I’m passing some raw data and getting it returned from GPT to answer some questions in a formatted manner. there are some recursive checks where I’m checking the response and prompting again to fix if there are any issues in the same message session (similar to a chatbot) via the program. As it is not time critical, I was wondering whether it is possible to use Batch API to reduce the cost and achieve same results?
Thanks in advance!

You write single independent API calls to an AI model with the chat completions endpoint when using batch for a 50% discount (or embeddings). They can have all the context of a conversation just like a chatbot’s API call, but typical use is just input/output on data.

Your checking would have to wait a day, as the period for fulfillment is up to 24 hours.