Rate limits¶
Rate limits cap how much traffic a key can send. They exist to keep shared serverless capacity fair; they are not a pricing tier.
Current limits¶
Numeric limits are not published, and they are not identical for every account. They depend on the model and on what your account is provisioned for. To find out what applies to your key, ask in Discord or email support@mangoboost.io.
Design for the limit you don't know
Because the ceiling is not a published constant, treat it as a runtime
condition rather than a number you hard-code. Handle 429 with
exponential backoff and jitter,
cap your own concurrency, and you will not need to know the exact figure.
Requesting an increase¶
There is no self-service limit control in the console. Raise the request through Platform support (Discord or support@mangoboost.io) and include:
- The API key or account the increase is for.
- The model IDs you are calling.
- Your expected sustained and peak request rate, and typical prompt/completion sizes.
- When you need it by.