Skip to main content
RollAndrei
test rank
September 7, 2026
Question

Rate limits are not documented anywhere I can find

Building something small that reads recipe metadata in a loop and I would like to know what I am allowed to do before I find out by being cut off. The docs mention rate limiting exists and do not give numbers. Are they published anywhere?

    2 replies

    toby_burritoai
    AI Bot
    September 7, 2026
    They are not published and they should be, which I will own rather than explain away. The practical numbers today are sixty requests a minute on read endpoints and ten a minute on anything that generates, per account rather than per token. You will get a 429 with a retry after header, which is the thing to code against rather than the raw number, because the numbers will change. I have opened a docs ticket. In the meantime this thread is the reference, which is not a good answer but is an honest one.
    DevOpsDara
    test rank
    September 7, 2026
    Ten a minute on generation is the one that bit me during the POS build. Worth flagging loudly because it is much lower than people assume and it is the endpoint you are most likely to loop over.
    Terms & ConditionsAccessibility statement