AI API status /

Is the OpenAI API reachable right now?

A live probe of api.openai.com from NetOkay’s edge, a test from your own browser, and the facts that separate an OpenAI outage from your own network.

In short: The card below shows the latest probe of api.openai.com from eight Cloudflare regions: one GET for /v1/models with a placeholder token. OpenAI answers in its own JSON: 401 invalid_api_key from most Cloudflare locations, or a 403 within milliseconds from locations in regions OpenAI does not serve. Either is OpenAI answering, not an outage. Then test from this browser: if NetOkay reaches the host and you do not, the problem is on your path.

Last updated

Where is api.openai.com reachable from?

Checking…

Loading the latest round from every region…

Loading the latest observation of api.openai.com…

From this browser

—

Not run. One request from your browser, sent only when you ask.

What this proves, and what it does not

Loading…

OpenAI specifics

API host
api.openai.com, HTTPS on port 443. Every model, embedding and file endpoint sits under /v1/ on this one host.
Streaming
Chat Completions and Responses stream as server-sent events. Each event is a data: line carrying a JSON chunk; Chat Completions ends the stream with data: [DONE], the Responses API ends with a typed response.completed event.
Reasoning pauses
Reasoning models (o-series and GPT‑5 reasoning modes) can stay silent for tens of seconds before the first visible token. A proxy that closes idle streams at 30 seconds cuts these responses off before any content arrives.
Official status
status.openai.com reports incidents per product. An incident there and a healthy probe here can both be true at the same time.
Support ticket header
Quote the x-request-id response header. The official Python and Node SDKs expose it as response._request_id, and on errors as request_id on the raised exception.
Client defaults
The official Python and Node SDKs wait up to 10 minutes per request and retry twice. A proxy that closes idle connections at 30 seconds fails long before the SDK would give up, so an SDK timeout is rarely what you see.

What OpenAI status codes mean

Code Sent by Meaning
403 API application Policy refusal in OpenAI’s own JSON: unsupported_country_region_territory for a region OpenAI does not serve, or a key without permission. Answered within milliseconds at the front door, so it proves DNS, TLS and reachability, not that your request would succeed.
401 API application Invalid or missing key, or the wrong project or organization header. The API application produced this answer.
429 API application Rate limit or exhausted quota; the JSON body says which.
500 / 503 API application Server-side error or overloaded. Retry with backoff.
Every row is an HTTP answer: the host was reached and the request was refused or served. None of them means api.openai.com is unreachable.

Reachable here but failing for you?

  1. Run the 30‑second idle stream from the network where your app fails. A reasoning pause looks like silence to a proxy: if the idle stream is cut, so is your response. Run the idle stream →
  2. Read the idle result. Closed during the silence points at a proxy, gateway or NAT idle limit under 30 seconds; start and done arriving together points at buffering. Fixes in the streaming guide →
  3. Compare environments if this browser passes but a container, server or agent fails, then run the CLI from the failing environment against api.openai.com. Compare environments →

OpenAI API status FAQ

Why does the probe show 401 or 403 when my key works?

The probe carries a fixed placeholder token that no account owns, so OpenAI answering 401 invalid_api_key is the expected result: the API itself refused the token. A 403 in OpenAI’s own JSON within milliseconds is a policy refusal, usually because the Cloudflare location that served you sits in a region OpenAI does not serve; a generic “Edge refused” block proves only DNS, TLS and the front door. None of these says anything about your key; if your own requests fail with 401, the key, project or organization header is the place to look.

My OpenAI stream stops mid‑response. Is that OpenAI?

Usually not. Run the 30‑second idle stream on the homepage. If it fails in your network but passes elsewhere, a proxy or load balancer on your path closes silent connections, which is exactly what a long reasoning pause looks like to it.

Does this page track OpenAI uptime over time?

No. It shows one recent observation cached for a minute per Cloudflare location. There is no history, no percentage and no alerting; use the official status page for incidents.