Current operational status of the Standard Compute API, verified by live requests against the production gateway when this page was loaded.
The indicator above is not hand-set. Loading this page sends real requests to the production API endpoints listed in the table, and the banner reflects what came back — including how long the gateway took to answer. If the gateway were down, this page would say so, because it could not learn anything else.
The same checks are available as JSON at standardcompute.com/api/status if you want to consume them programmatically or wire them into your own monitoring.
You do not have to take this page’s word for it. The models endpoint answers without authentication, so a single command verifies the API is up:
curl -i https://api.stdcmpt.com/v1/models
An HTTP 200 response confirms the gateway, routing layer, and model registry are serving requests. Authenticated endpoints respond to the same base URL with your API key.
We target 99.9% monthly uptime for the core API at https://api.stdcmpt.com/v1. This covers the infrastructure we operate directly — the gateway, authentication, key validation, and request routing.
Upstream model provider outages are outside our direct control. When a provider degrades, the router shifts eligible traffic to alternative providers for the same model class where possible, which is one of the practical benefits of not being tied to a single upstream.
When we detect a disruption, we work it until it is resolved and keep this page truthful throughout — the checks above fail automatically the moment the API stops answering. For significant incidents we publish a short post-incident summary on this page: what happened, the timeline, and what we changed.
If you are seeing symptoms this page does not reflect, email contact@standardcompute.com with your account email and a description. Reports go to the team that operates the gateway, not a ticket queue.
Planned maintenance is announced on this page in advance and scheduled in low-traffic windows, typically weekdays between 06:00 and 08:00 UTC. Most maintenance completes without dropping requests; when impact is expected, we state it here beforehand along with the expected duration.