OpenAI statusLiveFetched seconds agoSource: status.openai.com
Is OpenAI down?
Live status check · All systems operational
Live status pulled from status.openai.com, cached and revalidated every 60 seconds. Below: the components that can break, what to do during an incident, and how to keep your scheduled jobs running independently.
About OpenAI status
OpenAI runs the most-used AI APIs in production. Incidents typically affect a single product (ChatGPT, the API, Sora, the Playground) and a single model family; rate-limit pressure and elevated latency are far more common than full outages.
Components and surfaces that can break
- Surface 01
ChatGPT (web and apps).
- Surface 02
API endpoints (chat completions, embeddings, fine-tuning, batch).
- Surface 03
Sora, Playground, Operator, and the developer dashboard.
Recent OpenAI incidents
OpenAI maintains a public incident history. Read the latest entries to see whether you're hitting a known issue, and look at the post-mortems for context on how long similar incidents typically take to clear.
What to do if OpenAI is actually down
Separate elevated errors from a full outage
status.openai.com flags 'elevated errors' for partial degradation. That usually means a percentage of requests time out, not all of them. Backoff with jitter and you may not need to fail over.
Have a smaller model fallback
If gpt-5 is degraded, fall back to gpt-4o or a comparable Anthropic / Gemini model with the same prompt. Caching prompts client-side means you can swap providers in a single config change.
Pause scheduled batch jobs that aren't urgent
If a nightly classification cron is hammering the API during an incident, you're adding load while recovery is in progress. Pause non-urgent jobs and let real-time traffic recover first.
Keep firing through the incident
Schedule your own OpenAI health check with Crontap
Scheduled OpenAI jobs (nightly summarization, brand monitors, daily AI digests) are pure HTTP cron — fire a URL on your backend that calls the OpenAI API. Crontap retries on 5xx and 429, stores the response, and emails you when a batch job dies — useful both during OpenAI incidents and when your own code throws.
External HTTP cron hits your endpoint on the cadence you pick, stores every response, and emails you the moment a run fails. Pro schedules down to 1 minute; $3.25/mo annual flat for unlimited jobs.
Go deeper
OpenAI status: FAQ
- Where do I check OpenAI status?
- status.openai.com publishes component status for ChatGPT, the API, Sora, the Playground, and the dashboard. The history page has post-mortems for major incidents.
- Why am I getting 429 even though OpenAI looks operational?
- 429 is a rate-limit error, not an outage signal. Check your usage tier in the OpenAI dashboard — you may have hit your TPM (tokens per minute) or RPM (requests per minute) limit. The status page won't reflect that.
- Can I run a scheduled AI job that retries on OpenAI errors?
- Yes. Crontap retries 5xx and 429 responses with backoff before alerting. The job hits your wrapper endpoint, your endpoint calls OpenAI, and a transient API blip doesn't fail the whole run. See the OpenAI scheduled jobs sentiment pipeline post for a concrete example.