ChatGPT is down. Or crawling. And you have work to do. Why is ChatGPT slow or down and what to do about it? This is the emergency playbook — diagnose in 60 seconds, work around immediately, and prevent future disruption. Bookmark this one; you’ll need it eventually.
Why is ChatGPT slow or down and what to do?
The 60-second diagnosis
- Check the OpenAI status page — the single source of truth. Green across the board? It’s probably you. Red/yellow on ChatGPT? It’s them — skip to workarounds.
- Try a fresh chat with “hi” — works? Your old conversation was the problem. Fails? Systemic.
- Try a different device/network — phone on mobile data is the fastest isolation test. Works there? Your network or browser.
Sixty seconds, and you know which of the three worlds you’re in: their outage, your setup, or a specific conversation.
If it’s them (outage): the workaround kit
- Switch to a backup AI immediately. Don’t wait and refresh — in 2026 there’s no excuse for single-AI dependency. Keep 2-3 AI services bookmarked; your prompting skills transfer instantly. The outage is also a good time to evaluate whether your backup should become your primary.
- Use the API if you have access — API and chat outages don’t always coincide; one may work while the other doesn’t.
- Try the mobile app — sometimes different infrastructure, sometimes works when web doesn’t.
- Do the non-AI work first: triage your task list — what actually needs AI right now versus what you can do meanwhile? Outages reveal how much of your workflow is AI-dependent; note it for resilience planning.
- Check community trackers for ETA estimates — official pages give status, communities give “it’s back for me” signals.
If it’s you: the fast fixes
Why ChatGPT is slow today is the deep guide; the emergency version: hard refresh, incognito window (rules out extensions/cache in 10 seconds), different browser, disable VPN, restart router, try mobile data. Work the list fast — most “it’s down” reports are local issues with 2-minute fixes.
If it’s the conversation: rescue it
Long or corrupted chats fail while new ones work. Rescue: ask the old chat for a summary (if it loads at all), start fresh with the summary pasted in, and move on. If the old chat won’t load period, reconstruct from your notes/exports. This is why periodic exports matter — the chat you can’t open is the chat you should have backed up.
Specific errors and their guides
- “Something went wrong” → fix something went wrong error (systematic sequence)
- Network errors → fix network error messages
- General slowness → why ChatGPT is slow today
- Official help → OpenAI help center
Building outage resilience (do this once, benefit forever)
- Maintain 2-3 AI accounts across providers — free tiers suffice for backup purposes. The 10-minute setup pays for itself at the first outage.
- Keep prompts portable: store your best prompts in your own files, not locked in one platform’s features. Platform-agnostic prompting = instant switching.
- Export important work regularly — don’t let critical outputs exist only in a chat on a service that’s currently down.
- Know your non-AI fallbacks for critical tasks — the search engine still works during AI outages, remarkably.
- For businesses: multi-provider AI strategy isn’t paranoia, it’s operations. If AI is in your critical path, single-provider is a risk decision — make it deliberately.
The mindset
The resilience audit (for teams)
If your team depends on AI daily, run this audit quarterly: single points of failure — which workflows die if one AI service goes down? (Be specific: “client reporting,” “code review,” “content pipeline.”) Backup coverage — for each critical workflow, is there a tested alternative? (Tested means someone actually used it, not “we have an account.”) Export health — is important work backed up outside the AI platform? Skill redundancy — can the team do critical tasks without AI, slower but correctly? (The embarrassing question with the important answer.) Communication plan — when AI goes down mid-deadline, who decides the workaround? The audit takes an hour and reveals whether your AI adoption is resilient or fragile. Fragile isn’t failure — it’s information. Harden the critical paths, accept the risk on the rest, and stop being surprised by outages.
The status page habit
Bookmark your AI provider’s status page today — before you need it. When something feels off, the status page answers in 10 seconds what 20 minutes of troubleshooting can’t: “is it me or is it them?” Most major providers maintain one (usually at status.provider.com or similar), with incident history and subscribe-for-updates options. Turn on notifications if offered — knowing about an outage from the status page beats discovering it mid-deadline. This tiny habit eliminates the most common time-waster in AI troubleshooting: fixing your setup for a problem that was never yours.
Outages feel personal but aren’t — every cloud service has them, and AI services are still young infrastructure. The professionals aren’t the ones who never face downtime; they’re the ones with a backup tab open. Diagnose in 60 seconds, switch without drama, and use the downtime to build the resilience you should have built before. Then, when it’s back, export that important chat you’ve been meaning to back up.
