OpenAI status page explained — how to read it

You landed on status.openai.com and see a wall of green checkmarks, yellow warnings, or red outage bars—but you're not sure which component actually affects you. Here's how to read OpenAI's status page in 30 seconds: if

OpenAI status page explained — how to read it

You landed on status.openai.com and see a wall of green checkmarks, yellow warnings, or red outage bars—but you're not sure which component actually affects you. Here's how to read OpenAI's status page in 30 seconds: if the component you're using shows green, the problem is likely on your end; if it's yellow or red, OpenAI knows about it.

The live status monitor you're probably ignoring

OpenAI publishes real-time service health at status.openai.com, but most people don't know how to decode what they're seeing. Our live OpenAI status mirror on this site pulls that same data every 60 seconds, so you get identical information without leaving this page. The status page breaks OpenAI's infrastructure into specific components—ChatGPT Web, API Gateway, Sora, DALL·E, and more. Each component shows one of four states: operational (green), degraded performance (yellow), partial outage (orange), or major outage (red).

ChatGPT Web refers to the browser and mobile app interface most people use daily. If this shows red and you can't load chatgpt.com, that's a confirmed outage—not your internet. API Gateway covers developers hitting platform.openai.com endpoints for GPT-4, embeddings, fine-tuning, and batch jobs. A yellow warning here might mean 20% of API requests are timing out, while green means full availability. Sora, DALL·E Image Generation, and Whisper Speech-to-Text each have separate status rows. If you're trying to generate an image and DALL·E shows operational but ChatGPT Web shows degraded, the bottleneck is the web interface, not the model itself.

The status page also displays historical uptime—click any component to see a 90-day calendar with color-coded incident bars. If you notice ChatGPT was down three times last month between 2-4 AM UTC, you've learned peak maintenance windows.

What those yellow warnings actually mean

"Degraded performance" is vague on purpose—it covers everything from 5% slower response times to 50% of requests failing. OpenAI doesn't publish numeric thresholds, but you can infer severity from the incident description. "We are investigating reports of increased latency" usually means they've seen a spike in user complaints but haven't identified root cause. "Elevated error rates on ChatGPT Web" means a measurable percentage of logins or messages are failing. "Monitoring a fix" means they've deployed a patch and are watching metrics to confirm it worked.

If you see yellow but your experience feels normal, you're probably in the unaffected percentage. If you see green but ChatGPT isn't working, check our ChatGPT not working — troubleshooting guide—the issue is likely browser cache, cookies for chatgpt.com, extensions blocking WebSocket connections, or your network filtering OpenAI domains.

Component dependencies you need to know

API Gateway going down doesn't kill ChatGPT Web—they run on separate infrastructure. But ChatGPT Web *does* depend on API Gateway for model access, so a partial API outage can cause sporadic ChatGPT failures even when the Web component shows green. Similarly, if ChatGPT Web is operational but Authentication shows degraded, you'll load chatgpt.com but fail at login.

For developers, OpenAI API error 429 and other API errors often coincide with yellow API Gateway status. Error 503 (service unavailable) during a red outage is expected—retry with exponential backoff starting at 1 second, doubling to 2, 4, 8, 16 seconds. Error 429 (rate limit) during green status means you've hit your tier quota, not an outage.

How to verify what you're reading

Cross-reference OpenAI's status page with third-party monitors like Downdetector or X/Twitter search for "ChatGPT down". If OpenAI shows green but thousands of people are reporting failures in the past 10 minutes, the status page is lagging—manual incident creation can take 5-15 minutes. Conversely, if OpenAI shows red but you can use ChatGPT fine, you're in an unaffected region or the fix just deployed.

Test in incognito mode at chatgpt.com before assuming an outage. Clear cookies for chatgpt.com specifically—Chrome: Settings > Privacy and security > Cookies and other site data > See all site data and permissions, search "chatgpt.com", click the trash icon. Disable browser extensions like ad blockers, VPNs, and privacy tools—these frequently break WebSocket connections required for streaming responses. Try the mobile app if the desktop site fails, or vice versa. Our live OpenAI status updates every 60 seconds, so refresh it to catch status changes in near real-time.

Questions people actually ask

Q: Is ChatGPT down right now?

A: Check the live OpenAI status monitor on this page—it pulls from OpenAI's official feed every 60 seconds. If ChatGPT Web shows green, the problem is likely your browser cache, extensions, or network.

Q: Why does the status page show green when I can't connect?

A: Status updates can lag 5-15 minutes behind actual outages, or you're hitting a localized network issue. Test in incognito mode and disable extensions first.

Q: What's the difference between "investigating" and "monitoring"?

A: "Investigating" means OpenAI hasn't identified the root cause yet. "Monitoring" means they've deployed a fix and are confirming it resolved the issue.

Q: Do I need to contact support during an outage?

A: No—OpenAI already knows. Submitting tickets won't speed up resolution. See how to contact OpenAI support only for account-specific issues unrelated to the outage.

What to remember

  • ChatGPT Web and API Gateway are separate components—one can fail while the other works fine
  • Yellow warnings mean partial degradation, not total failure—you might be in the unaffected percentage
  • Status updates lag real incidents by 5-15 minutes during fast-moving outages
  • Green status with continued failures points to browser cache, extensions, cookies, or network filtering on your end
  • Historical incident calendars reveal maintenance patterns and recurring downtime windows

---

*openai-support.com is an independent resource, not affiliated with OpenAI Inc.*

Related help