Evaluation
What evaluation access includes
Your evaluation gives you temporary, scoped access to Aimable as a gateway so you can validate it against your own workload before committing.
The details below in [brackets] are set per evaluation. Replace them with the values from your evaluation welcome email before sharing this page externally.
- Base URL:
https://<your-tenant>.aimable.ai/api— your<your-tenant>subdomain is in your welcome email. - Credentials: an API key (
ak.<key_id>.<secret>) and at least one space id, provided with your welcome email. - Available models:
[list the models enabled for evaluation] - Duration:
[evaluation period, e.g. 30 days] - Support contact:
[support email / channel]
A suggested evaluation path
- Connect. Make a first call against
/v1/chat/completionsfollowing Getting Started. Confirm auth, the space header, and a successful completion. - Stream. Repeat with
"stream": trueand confirm Server-Sent Events arrive incrementally. - Swap your client. Point your existing OpenAI-compatible client at the base URL and add the
X-API-KeyandX-Aimable-Space-Idheaders — verify your current code works unchanged. - Provision another space. Create a second space and a service principal + key following Provisioning. Confirm the two spaces are isolated.
- Try governance. Send a request through
/v1/spaces/{space_id}/responseswith PII pseudonymization enabled and inspect the result andX-Routing-Details. - Check metering. After some calls, read usage metrics per space and confirm cost and token attribution match what you expect.
Current limits to be aware of
- Rate limiting: the governed
/responsesendpoint enforces 30 requests/minute per principal and returns429withRetry-After. Build retry-with-backoff. - Quotas: hard per-space or per-tenant spend/token quotas are not yet enforced. During evaluation, track spend via the usage metrics.
- Metering: usage is read through the observability metrics endpoints, which require your tenant's observability to be configured. There is no separate billing API.
Getting help
If a call behaves unexpectedly, capture the X-Request-Id response header and include it when you contact [support contact]. It lets us trace the exact request.
Next steps
When you are ready to move beyond evaluation, contact [support contact] to discuss production access, model availability, and commercial terms.