Ngamia publishes default limits for authentication and chat-completion traffic. Treat the numbers below as deployment defaults rather than a permanent contract; voice and other media routes may also be constrained by service, reverse-proxy, or platform limits.
Default limits
| Surface | Limit key | Default |
|---|---|---|
/v1/auth/* | Client IP address | 20 requests per minute |
POST /v1/chat/completions | API-key id | 100 requests per minute |
Other routes do not have a separate documented request-per-minute default in this portal. Account, service, and deployment limits may change over time, so clients should handle 429 responses on every route.
Use an API key for production gateway traffic. JWT-authenticated calls are intended for authenticated account sessions, while API-key traffic is the stable server-to-server integration path.
Handle a 429
A rate-limited request returns the standard Ngamia envelope with error_code: "rate_limited":
{
"status": "error",
"code": 429,
"error_code": "rate_limited",
"message": "too many requests, slow down",
"request_id": "req_..."
}If a response includes Retry-After, respect it. Otherwise use your own exponential backoff with jitter:
- Wait briefly after the first
429. - Increase the delay after each consecutive
429. - Add random jitter so multiple workers do not retry together.
- Stop after a bounded number of attempts and surface the failure.
Do not retry validation, authentication, permission, or not-found errors unchanged. See Errors and response format for the full recovery table.
Other limits to plan for
| Constraint | Default |
|---|---|
| Standard JSON request body | Keep requests small; the effective limit is the lower of the service and deployment limits. |
| Direct transcription audio | 25 MiB service target; a lower edge or deployment cap may apply. |
| Speech input | 100,000 Unicode characters; keep the JSON body within deployment limits. |
| Active API keys per account | 50 |
| Workspaces per account | 20 |
List endpoint limit | 50 by default, capped at 200 |
Notifications list limit | 20 by default, capped at 100 |
See API keys, Profile & workspaces, and Notifications for resource-specific details.