Is OpenAI (ChatGPT) down? Server status
OpenAI (ChatGPT) posted 25 incidents on its official status page in the last 30 days, usually resolved within 1.7h. Updated every 30 minutes.
Now: Operational
As of Nov 8, 20:45 UTC, the official status of OpenAI (ChatGPT) is operational.
Recent trend
4 incidents in the last 7 days — within the usual range.
Incident forecast
- Within 24 hours
- 55%
- Within 72 hours
- 90%
We'll notify you on the site and by email when OpenAI (ChatGPT) goes down or recovers.
Last 7 days11/1 – 11/8
- Incidents
- 4
- Major+
- 0
- Days with issues
- 4
- Median recovery
- 1.7h
- Most recovered by
- —
The line above is the measured API success rate; the band below shows reported outages.
- Core feature outageOutagePartial outageBrief outage
- Non-core issue
- Normal
- No record
API success: 24 hours 100%, 7 days 100%. (measured since 11/6)
API success is the share of requests OpenRouter sent directly to each company’s API that succeeded. It measures the API, not the app, independently of status notices.
Last 90 days8/11 – 11/8
- Incidents
- 54
- Per week
- 7.1
- Major+
- 8
- Days with issues
- 34
- Median recovery
- 1.6h
- Most recovered by
- 8.3h
* Data from before 53 days ago has not been collected yet
- Operational
- Minor issue
- Major issue
- Critical outage
- No record
Outage hours, last 12 weeks8/16 – 11/8
- Core outage, weekly avg
- 18h
- Non-core issues, weekly avg
- 15.8h
- Core feature outage
- Non-core issue
- No record
Incident cause analysis
What was affected
- App & web19
- API13
- Sign-in & billing4
- Other / unclear3
- Image, audio & video1
Cause (as far as the notice says)
- Cause not stated38
- Deploy or config change3
- Core use (chat, API and the like) was visibly affected in 26 of 41 labeled incidents.
- "App & web" problems repeated 4 times in the last 14 days.
Recent incidentsTitles as published
- Nov 8, 15:11 UTC
Image, audio & videoElevated errors with image generationCore feature outage・lasted 1.8h
- Nov 5, 15:25 UTC
ChatGPT Work and Dots ErrorsCore feature outage・lasted 1.7h
- Nov 4, 22:38 UTC
App & webElevated error rates across paid ChatGPT plansCore feature outage・lasted 11m
- Nov 3, 00:46 UTC
App & webFile upload errors in ChatGPTNon-core issue・lasted 4.9h
- Oct 30, 19:38 UTC
App & webElevated errors with search in Dots, ChatGPT Work, and certain ChatGPT thinking experiencesNon-core issue・lasted 1.6h
- Oct 30, 14:35 UTC
APIElevated errors for GPT Image 2 API image editsCore feature outage・lasted 4h
- Oct 28, 03:39 UTC
App & webDegraded Performace on ChatGPT WorkNon-core issue
- Oct 27, 15:02 UTC
Other / unclear6.1 Sol Degraded Model PerformanceCore feature outage・lasted 78m
- Oct 26, 18:19 UTC
APIElevated error rates for Ads POST requestsNon-core issue・lasted 11m
- Oct 23, 22:52 UTC
App & webDegraded experience with ChatGPT app connections using CIMDCore feature outage・lasted 50m
- Oct 22, 19:36 UTC
Sign-in & billingNew Ad Accounts failing to provision adminsNon-core issue・lasted 4.5h
- Oct 22, 18:51 UTC
App & webDegraded Performance of Dots on EnterpriseCore feature outage・lasted 2.1h
- Oct 22, 02:18 UTC
APIElevated ChatGPT conversation and Responses API errorsCore feature outage・lasted 6.5h
- Oct 21, 15:34 UTC
APIElevated error rates for gpt-5.6-luna API requestsCore feature outage・lasted 69m
- Oct 21, 13:51 UTC
APIElevated FedRAMP API error rates for Astra and SolCore feature outage・lasted 3.2h
OpenAI (ChatGPT) news
More news- •Former OpenAI Engineer Pleads Not GuiltyNovember 8, 2026
- •OpenAI Publishes GPT Image Prompting GuideNovember 8, 2026
- •OpenAI Pledges to Pay Data Center CostsNovember 8, 2026
- •OpenAI Sets Data Center Trust PoliciesNovember 8, 2026
- •OpenAI Hires Dan Driscoll for Security RoleNovember 8, 2026
- •OpenAI Hires Claude Code Lead Cat WuNovember 8, 2026
- •OpenAI Presses Ahead With Data Center InvestmentNovember 7, 2026
- •Voters Reject Data Center Expansion in 10 StatesNovember 7, 2026
- •ING Offers $500 Million in Data Center LoansNovember 7, 2026
- •Tech Executives Address Midterm Sell-Off FearsNovember 7, 2026
OpenAI (ChatGPT) FAQ
- Is OpenAI (ChatGPT) down right now?
- As of Nov 8, 20:45 UTC, the official status of OpenAI (ChatGPT) is operational.
- How long does OpenAI (ChatGPT) usually take to recover from an outage?
- In the last 53 days of records, the median recovery time for OpenAI (ChatGPT) incidents was 1.6h, and 9 out of 10 were resolved within 8.3h.
- How often does OpenAI (ChatGPT) have incidents?
- Over the last 30 days, 25 incidents were recorded on the official status page, an average of 5.8 per week.
- What can I use instead when OpenAI (ChatGPT) is down?
- You can use Claude, Gemini, Grok, DeepSeek and others instead of OpenAI (ChatGPT). You can check the current status of each one on this site.
- What if OpenAI (ChatGPT) shows errors but the official page says it's fine?
- Problems limited to some regions or features may be posted late or not at all. Try reloading, signing out and back in, and turning off a VPN. If it still fails, check whether you have hit a usage limit.
Model response speed trend
Output speed and time to first token measured once a day through OpenRouter over the last 30 days. “Usual” is the median of the previous 14 days; a model is marked slower or faster only when it moves beyond its own normal variation. Speed you experience may differ by region and time of day.
- GPT-6 LunaUsual
Recent 70 tok/s · first token 2.4s
- GPT-5.6 LunaUsual
Recent 68 tok/s · first token 1.7s
- GPT OSS 120BUsual
Recent 153 tok/s · first token 0.3s
- GPT-6.1 SolUsual
Recent 45 tok/s · first token 3.7s
Sources and methodology
- We collect incidents from OpenAI (ChatGPT)'s official status page every 30 minutes and keep them. Past incidents removed from the official page stay in this record.
- Incident counts and recovery times cover only the period since we started recording. When that period is short, we show "53 days".
- Affected features and incident causes are classified by AI from the incident notices.
- The outage probability is a reference estimate from the last 60 days of records (newer weighted more) and the average across all services.
- Maintenance notices are not counted as incidents. Providers post by different standards, so more notices don't necessarily mean less stable.