A rate-limiting algorithm where capacity refills continuously up to a cap; a request is allowed only if enough tokens are available, and is denied (not queued) otherwise.
Continue to AI University →