What to do during a ChatGPT outage

You're staring at an error message, the chat won't load, or ChatGPT just vanished mid-conversation. Before you panic or restart your entire workflow, you need to know if this is a widespread outage or something you can fix in two minutes. Here's how to check within 30 seconds: visit the live OpenAI status monitor on this site, which updates every 60 seconds from OpenAI's official feed, or check status.openai.com directly for component-level reports.

How to check if OpenAI is actually down

Start with this site's live OpenAI status page. It pulls directly from OpenAI's infrastructure status and refreshes every minute, showing you exactly which services are degraded or offline — ChatGPT web, mobile app, API, Sora, DALL·E, or login systems. If you see red indicators next to ChatGPT, you're looking at a confirmed outage.

Next, go to status.openai.com and check the component breakdown. OpenAI separates their services into specific systems: chatgpt.com (the web interface), ChatGPT mobile apps (iOS and Android), API endpoints, authentication services, and others. An outage might hit only the web interface while the mobile app works fine, or vice versa. Look at the incident timeline — OpenAI posts updates every 15-30 minutes during major outages, including estimated resolution times.

Third-party reports confirm widespread problems. Check Twitter/X for "@OpenAI down" or "@ChatGPT outage" — if hundreds of people are posting in the last 5-10 minutes, it's not just you. Downdetector.com shows real-time user reports with a spike graph. If you see a sudden vertical line in the last hour, that's a confirmed mass outage.

Is it OpenAI or is it you

If status monitors show green but ChatGPT won't work for you, isolate the problem with these exact steps:

Open an incognito/private browser window and try chatgpt.com again. This bypasses cached data and extensions. If it works here, you've got a local browser issue.

Clear chatgpt.com-specific cookies and cache. In Chrome: Settings → Privacy and security → Delete browsing data → check "Cookies" and "Cached images" → Time range: "All time" → enter "chatgpt.com" in the search box → Delete data. In Safari: Preferences → Privacy → Manage Website Data → search "chatgpt.com" → Remove.

Disable all browser extensions temporarily. Ad blockers, privacy tools, and VPNs cause 60% of individual ChatGPT access problems. Turn them off, refresh chatgpt.com, and see if the problem disappears.

Switch networks. If you're on corporate WiFi or a VPN, disconnect and try mobile data or a home network. Some workplace firewalls block ChatGPT's WebSocket connections required for streaming responses.

Try a different platform. If the web version fails, open the ChatGPT iOS or Android app. If the app fails, try the web version on a different device. Outages sometimes affect only one platform.

For detailed troubleshooting beyond these steps, see our guide on ChatGPT not working — troubleshooting.

What to do during an outage

Once you've confirmed it's a genuine OpenAI outage, follow these numbered steps:

Stop retrying immediately. Hammering the refresh button or resending API requests makes the problem worse for everyone. OpenAI's systems need breathing room to recover.

Check the live status monitor every 15 minutes for incident updates. OpenAI posts progress reports during major outages — "investigating", "identified", "monitoring", "resolved" — so you know when to try again.

For API users: implement exponential backoff. If you're running automated requests, add retry logic with delays that double each time: wait 1 second, then 2, then 4, then 8, up to a 60-second maximum. Set a retry budget of 5 attempts maximum before giving up for 10 minutes.

Queue your work instead of forcing it. Write your prompts in a text file, save your API calls to a job queue, or draft content offline. When service returns, you'll resume immediately instead of scrambling to remember what you were doing.

Don't make desperate account changes. Never delete your ChatGPT account, cancel your Plus subscription, or create a new account during an outage. These actions won't restore service and create unnecessary problems when systems recover.

Don't contact support about outages. OpenAI's support team can't fix infrastructure problems faster, and they're already swamped during incidents. Support is for account-specific issues, billing problems, or policy questions — not downtime reports.

Questions people actually ask

Q: Is ChatGPT down right now?

A: Check the live OpenAI status monitor at the top of this site — it updates every 60 seconds. If all indicators show green, the issue is likely on your end.

Q: How long do ChatGPT outages usually last?

A: Most outages resolve in 15-45 minutes. Severe incidents affecting multiple services can take 2-4 hours. OpenAI posts estimated resolution times on status.openai.com during major events.

Q: Will I get refunded for ChatGPT Plus downtime?

A: OpenAI doesn't automatically issue credits for brief outages. For extended downtime exceeding 24 hours, you can request a prorated refund through their support system.

Q: Can I use the API during a ChatGPT web outage?

A: Sometimes yes — check status.openai.com to see if API services show operational while chatgpt.com is degraded. The systems run separately.

What to remember

Use this site's live status monitor or status.openai.com as your first stop — they update before social media reports
Try incognito mode and disable extensions before assuming it's an outage
Stop retrying requests during confirmed outages — you're making recovery slower
Implement exponential backoff and retry budgets for API integrations
Never delete accounts or cancel subscriptions during temporary downtime

Related help

---

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