Is ChatGPT down right now?

TL;DR#

• Check status.openai.com first. It's the fastest official signal.
• If the status page is green and ChatGPT still won't work, the problem is probably on your end: rate limits, cache, VPN, or browser.
• The API can go down while the web app is fine. Developers should check that component on its own.
• statuspy.com checks ChatGPT independently, around the clock, along with 126 other services. Useful when OpenAI's own page is behind reality.

How to check ChatGPT's current status#

Go to the source first.

status.openai.com is OpenAI's official status page. It lists every component (ChatGPT, the API, Sora, image generation) with live indicators and a full incident history.

If everything there is green and ChatGPT still won't load, the outage isn't global. Skip to the next section.

Two other signals worth having open:

• statuspy.com runs its own checks on ChatGPT down or not continuously. It doesn't depend on OpenAI's self-reported data, so it catches incidents the official page is slow to acknowledge.
• Downdetector (downdetector.com/status/openai) collects user reports. They spike fast, often before OpenAI updates its own page.

One caveat about the official page: it has lagged behind real incidents. During the June 10 to 11, 2025 outage, the biggest of that year, people were reporting failures for hours before the page showed the full scope. That gap is the whole argument for keeping a third-party monitor bookmarked.

The status page is green. So why isn't ChatGPT working for you?#

This is the common case, and the fix is usually quick.

You've hit a rate limit#

ChatGPT caps usage by plan. Free accounts get roughly 10 GPT-4o messages per 3 hours before being pushed down to a lighter model. Plus accounts get around 80 per 3 hours.

When you hit the cap, you rarely get a clear error. ChatGPT slows down, quietly switches models, or shows a vague "something went wrong." Wait 30 to 60 minutes and try again. On the API side, check your usage dashboard at platform.openai.com/usage.

Your cache or session is stale#

A stale cache or a corrupted session cookie makes ChatGPT look broken when the service is fine.

Try these in order:

1. Hard-refresh the page (Ctrl+Shift+R on Windows, Cmd+Shift+R on Mac).
2. Open an incognito or private window and load chat.openai.com.
3. Clear cookies and site data for openai.com specifically.
4. Try a different browser.

If it works in incognito, the problem is your session or an extension. Ad blockers and privacy tools are the usual culprits.

Your network or VPN is in the way#

ChatGPT is geo-restricted in some countries and blocked on plenty of corporate and university networks. If you're on a VPN, turn it off and retry. OpenAI flags some VPN exit nodes.

If you're in a restricted region and need the VPN to get in at all, switch to a different server location. DNS can also fail silently, so try 1.1.1.1 (Cloudflare) or 8.8.8.8 (Google) before you give up.

The API is down even though the web app is fine#

This one catches developers out. The API and the web app run on separate infrastructure. chat.openai.com can load perfectly while your API calls fail, and the reverse happens too.

Check the API component on status.openai.com separately from the ChatGPT component. It's its own line item. If your app is throwing errors and the web UI looks healthy, that's your answer.

How reliable is ChatGPT, really?#

Reliable most of the time, with real outages. Some ran for hours. One ran most of a day.

The significant incidents of 2025:

June 2025 is the one that made global headlines: millions of users, ChatGPT slow and then unreachable, and long stretches with the API down.

statuspy.com keeps 90-day uptime history for ChatGPT and 126 other services, updated continuously. If you want the current uptime percentage and a recent incident log that doesn't come from OpenAI, that's where to look.

For most people ChatGPT's reliability is good. If your workflow depends on it, a second monitor costs you nothing.

What to do while ChatGPT is down#

Switch to something else#

• Claude (Anthropic): strong reasoning and a long context window. Usually the first stop for people doing writing or analysis work.
• Gemini (Google): tied into Google Workspace, so it's the easy switch if you already live there.
• Perplexity: search-native, good for anything that needs live web results.

All three have free tiers. Claude and Gemini both have APIs with trial credits if you need to reroute code, not just chats.

Read the status page properly#

At status.openai.com, look at three things:

1. The status banner at the top.
2. The component list, where ChatGPT, the API, and the rest are tracked separately.
3. The incident history at the bottom for the latest updates.

You can subscribe to updates by email or RSS from that same page, which is the easiest way to find out the moment an incident is closed.

Watch OpenAI on X#

@OpenAI posts incident acknowledgments and resolutions, sometimes ahead of the status page. During the June 2025 outage it was one of the first official channels to confirm the problem.

For API incidents, watch @OpenAIDevs. Developer-facing updates land there first.

Sources#

• status.openai.com: official status page, component-level breakdown and incident history
• status.openai.com/history: full historical incident log
• help.openai.com: OpenAI Help Center for account and billing issues
• x.com/OpenAI: real-time incident updates
• x.com/OpenAIDevs: API-specific updates
• downdetector.com/status/openai: user-reported outage signals, useful for early detection