Status
API status
Read live from GET /v1/status each time this page loads.
Checking status…
The panel
Model lanes
A lane is in the panel once its upstream key is in place. Until the panel is ready, chat requests return 503 with "not live yet".
Loading the panel…
Endpoints
What is served
| Endpoint | Auth | Purpose |
|---|---|---|
| POST /v1/chat/completions | API key | Chat Completions, streaming supported |
| GET /v1/models | API key | The model ids you can call |
| GET /v1/usage | API key | This key's month spend, cap, remaining and per-model breakdown |
| GET /v1/pricing | Public | Current prices per 1M tokens |
| GET /v1/status | Public | Live flag and lane readiness |
Base URL https://api.failure.fail/v1, also served at https://failure.fail/v1.
Check it yourself
No key needed
Status and pricing are public. Poll status from your own monitoring before you fail over.
curl https://api.failure.fail/v1/status
# live: true once the panel is ready
# lanes: one entry per model, with a ready flag
Reading errors
When something is down
503
service_unavailable
Not live yet, or the router is restarting. Retry with backoff.
402
insufficient_quota
Not an outage: the key has reached its monthly spend cap.
401
invalid_api_key
Not an outage: the key is wrong or revoked.