Introduction
Rate limits
How throughput and quotas apply to myt AI Cloud requests.
Rate limiting is enforced per organisation and per model according to your deployment settings (requests per minute, tokens per minute, and concurrency). The portal surfaces effective limits for your projects where configured.
Practical guidance
- Retry with backoff on
429responses using jitter - Cache stable prompts where possible to reduce duplicate spend
- Stream long outputs to improve time-to-first-token UX without changing limits
Platform policy
Login and unauthenticated routes may apply per-IP throttles; authenticated API traffic is attributed to your API key and organisation. Refer to your operator for exact numbers.