Is OpenAI API down?
api.openai.com · AI tools · official status page
OpenAI API has recorded 100.0% uptime over the last 5 days, with no outage. It answers our checks in 230 ms on average.
No problems detected
OpenAI API answered our last automated check normally, 11 minutes ago. If it is not working for you, the problem is most likely local - Your connection, your DNS resolver, a cached page, or an extension.
OpenAI API problem reports in the last 24 hours
Counted in 15-minute buckets from reports submitted on this site. This is a user signal - A tall bar means people are complaining, not that our check failed.
No reports in the last 24 hours.
If OpenAI API is not working for you, tell us and the chart starts here.
Response time from our automated checks
The last 24 checks of https://api.openai.com/v1/models from our network. Bar height is the response time; the tallest bar is the slowest.
- Answered normally
- Slow to answer
- No usable answer
Having a problem with OpenAI API?
Tell us what is not working. Reports go straight onto the 24-hour chart above and help other people see a problem before any company admits to one. 0 reports in the last 24 hours.
About OpenAI API outages
The OpenAI API is a dependency inside thousands of products that never mention it, so an API incident surfaces as unrelated apps losing their AI features simultaneously - A support chatbot that stops answering, a writing tool that times out, a code assistant that returns errors.
Its availability is separate from the ChatGPT consumer product, and the status page reports them apart. The distinctive error patterns are 429 for rate limits and quota exhaustion, 500 and 503 for service problems, and 529 for overload. Per-model availability varies: A newly released model is far more likely to be capacity-constrained than an established one, so switching model in a client's configuration is often a faster fix than waiting. Latency degradation without errors is the most common incident type and is easy to miss because requests still succeed, just much more slowly.
What to do if OpenAI API is not working for you
Is it them or you?
Our last check reached api.openai.com normally, from a different network, resolver and route than yours. If it is broken for you, the fault is most likely local or too narrow for a single probe to see. Run the check again now to confirm.
The first thing worth trying
Distinguish slow from down before reporting. A response that takes ninety seconds is a degradation. Knowing which one you are seeing changes what you should do.
How ai tools services typically fail
Requests are accepted and take far longer than usual, or return a rate-limit error. Answers may truncate mid-sentence. The interface stays perfectly responsive, which is why this reads as a slow service rather than a broken one. The rest of the steps, and why this category breaks the way it does, are on our ai tools status page, with more background in 429 Too Many Requests explained.
The link below unfurls into a live card showing the verdict, so whoever you send it to sees the evidence rather than your word for it.
Put a live badge on your own site
An SVG that always shows OpenAI API’s current state, updated from our checks. Works in a README, a status dashboard or a footer.
Free, no account, and it will keep working. We ask nothing in return beyond the link back.
OpenAI API measured against its peers
How it compares with other ai tools services
Over the last 30 days OpenAI API ranks 4 of 5 ai tools services we monitor, measured on recorded uptime. The category averages 100.0%, and OpenAI API sits at 100.0%. The steadiest in this group is ChatGPT.
On speed, OpenAI API answers our checks in about 230 ms against a category average of 195 ms - Slower than its peers. Response time is not availability, but a service drifting well above its own normal is often the first sign of trouble.
What the last 24 checks looked like
Across the most recent 24 checks, the quickest response was 133 ms and the slowest 1200 ms, against a 30-day average of 230 ms. The chart above plots each one, so a widening spread is visible before it becomes an outage.
All figures are our own measurements, taken every 20 minutes from one network. What that can and cannot see is set out in the methodology, and every ai tools service is listed on the category page.
The server that answered Mostly hidden
- IP address
- 162.159.140.245 and 3 others
- Server
- Hidden until you sign in
- Powered by
- Hidden until you sign in
- Content type
- Hidden until you sign in
- Declared size
- Hidden until you sign in
- Transport and HSTS
- Hidden until you sign in
Five more fields about the server behind api.openai.com.
Free, one click with Google, and nothing is posted or shared. We ask because the lookups behind this cost us real requests to other people’s servers.
Read from our last sweep of api.openai.com, not requested when you opened this page.
Certificate Mostly hidden
- Status
- Valid
- Issued to
- Hidden until you sign in
- Issued by
- Hidden until you sign in
- Valid from
- Hidden until you sign in
- Expires
- Hidden until you sign in
- Protocol and cipher
- Hidden until you sign in
Who issued it, what it covers and exactly when it expires.
Free, one click with Google, and nothing is posted or shared. We ask because the lookups behind this cost us real requests to other people’s servers.
An expired certificate stops browsers loading a site entirely while the server itself stays perfectly healthy, which is why we track it beside the uptime figure. Check it yourself.
OpenAI API status: Common questions
Is OpenAI API down right now?
OpenAI API is up and responding normally. This page is refreshed by an automated check that runs every 20 minutes from an independent network, and by reports from visitors.
Why can I not reach OpenAI API when you say it is up?
Because an outage can be partial. Our check runs from one network and asks for one page. Your problem may be regional, may affect only the mobile app, or may be local to your connection, DNS resolver or browser cache.
How reliable is OpenAI API?
OpenAI API has been reachable for 100.0% of the last 5 days we have measured, with 0 seconds of recorded downtime.
Does OpenAI API have an official status page?
Yes - OpenAI API publishes one at https://status.openai.com/. It is the authoritative source for confirmed incidents, though official pages usually update after user reports have already peaked.
OpenAI API discussion
Useful comments say what broke, roughly where you are and roughly when it started - That is what turns a pile of complaints into something other people can act on. Links are shown as plain text and carry no SEO value, so there is nothing here worth spamming.