ChatGPT experienced a widespread outage starting at 3 a.m. ET on June 10, affecting users across all platforms for over 11 hours. OpenAI, the company behind ChatGPT, acknowledged the service disruption, which also impacted its Sora video generation platform and APIs, with the company working throughout the day to restore full functionality.
What you should know: The outage affected ChatGPT across all platforms, including the website and mobile applications.
- Users encountered error messages stating there was “an error in the message stream” or that “something seems to have gone wrong” when attempting to use the service.
- The disruption impacted ChatGPT, Sora text-to-video platform, and OpenAI’s APIs simultaneously.
Timeline of the outage: Reports began trickling in around 3 a.m. ET before escalating significantly by morning.
- User reports to Downdetector, a website that tracks service outages, spiked at approximately 5:30 a.m. ET, reaching over 1,000 reports at peak.
- At 9:07 a.m., OpenAI announced it had “identified the root cause of the issue and was working on fixing the problem.”
- By 10:54 a.m., the company reported “some recovery to its APIs” but warned that full recovery across all services “could take another few hours.”
- As of 2 p.m. ET—11 hours after the initial outage—OpenAI was still “monitoring the fix and working toward a full recovery.”
Alternative AI tools to consider: Several robust alternatives remain available during ChatGPT downtime.
- Google Gemini offers both free and paid versions for general questions and requests.
- Microsoft Copilot provides voice conversation capabilities, particularly when accessed directly through Windows.
- Other recommended alternatives include Anthropic’s Claude AI and Perplexity.
- According to the report, “the freebies should be able to handle just about any challenge you throw at them.”
Why this matters: The outage highlights the growing dependence on AI services and the need for backup options when primary tools fail.
Is ChatGPT still down for you? Here's when OpenAI expects to have it back up and running