Checks
Running…
Catalog & pricing
One row per model id seen in the catalog, the pricing endpoint or the deprecation history's upcoming removals. Prices are the pricing endpoint's raw numbers per unit; the endpoint returns no currency field.
| Model id | Provider | Task | Input | Output | Cached | Status |
|---|
Live contract
What the API actually returns, set against the errors reference. The no-key probe runs for everyone. The first-call runner uses your own key, forwarded once through this site's proxy and never stored.
No-key probe
| Request | Status | Content type | Shape | Body |
|---|
The gateway checks the key before routing: an unknown path answers the same 401 as a real one. So a no-key probe proves which error shape each route uses; it cannot prove which routes exist.
First-call runner your key
Response vs the docs
Release note draft
Built from the deprecation history and a diff against the catalog and pricing frozen on 17 Oct 2026 UTC. Scheduled removals first, then recent removals grouped by replacement, then catalog and price changes.
OpenAPI fragment & llms.txt
Two machine-readable artefacts the docs do not publish yet, generated from what the desk observed: the error schema as it behaves on the wire, and an llms.txt index built from the sitemap and catalog.
openapi-errors.yaml
llms.txt
What this is
A docs QA desk for an inference API. It reads Bitdeer AI's developer docs, the public pricing endpoint, the product page and the live API, then checks that they agree: the quickstart model exists, code samples use catalog models, deprecation replacements are real ids, pricing and catalog list the same models, error codes match production, and machine-readable docs exist. From the same data it drafts a release note, an OpenAPI error fragment and an llms.txt.
Built by Edward Tay for an application to Bitdeer's Technical Writer, MaaS role. It is not Bitdeer software and uses only public pages and endpoints.
Sources
- Docs: model catalog, deprecation history, errors, first request, every English guide in the sitemap.
- Pricing:
POST https://www.bitdeer.ai/api/model/v1/ListPublicModelPricings, shown on the pricing page. - Product page: AI inference.
- API:
https://api-inference.bitdeer.ai/v1, probed without a key.
Limits
- Model names on the product page are matched by vendor family and id suffix.
- The no-key probe shows the error shape per route; confirming that a route exists needs a key.
- Your API key goes from this page to this site's worker and on to api-inference.bitdeer.ai for one request. It is held in memory only.