Prices

Is the Cerebras API down?

As of 16:05 UTC: up (answered in 1.1 s from our vantage). Last 24 h: 100.0% up.

Last probe: (UTC), one request every 5 minutes from one vantage (werk-nl). Latest class: up: an expected answer (HTTP 401, 403 or 200) inside 10 seconds.

Uptime and latency

PeriodUptimep50 / p95 latencyProbes
24h100.00% 911 / 1415 ms 288
7dcollecting n/a 471 (23% of the period covered)
30dcollecting n/a 471 (6% of the period covered)

Uptime = (up + degraded) / (up + degraded + down) over the measured windows.

Incidents, last 30 days

No incident in the last 30 days: no run of 2 or more consecutive failed probes.

up = an expected status (401, 403 or 200) within 10 s; degraded = answered but slower than 3x the host's own 7-day p95, or a one-off 5xx; down = no answer or a repeated 5xx; unexpected = an unexpected status, not counted; probe_fault = our own network failed for nearly all hosts, not counted. An outage is claimed only after 2 or more consecutive down windows (one failed probe is a blip: it may be our own network).

What this does and does not prove

What this proves: from this vantage (one IPv4 network, one location) the provider's DNS, TCP, TLS and edge or gateway answered, and how fast. What it does NOT prove: that a model generates, that your key, quota or region works, that other regions or IP ranges are up, or that a failure we saw is global. A provider can be fully up for this probe while inference is failing, and the reverse. Treat a down window as 'unreachable from here', and use several sources (and the provider's own status page) for a claim.

Probe: one unauthenticated GET to api.cerebras.ai/v1/models. It identifies itself as InfraRabbitBot/1.0 (+https://infrarabbit.com/bot). How it is measured: method.

All AI API hosts, status right now · JSON: https://infrarabbit.com/prices/uptime/cerebras-api.json, daily history · MCP tool get_uptime at https://infrarabbit.com/mcp. CC BY 4.0.