ChatGPT Hit by Major Global Outage on April 20

ChatGPT went down for thousands of users worldwide today, with OpenAI’s status page confirming a partial outage that lasted roughly 90 minutes and knocked out core features across the platform.

What Happened

The problems started around 10:05 AM ET on April 20, 2026. Users reported blank screens, endless loading spinners, gateway timeouts, and 403 forbidden errors. Many could not log in at all, while others could open ChatGPT but found it unable to send or receive messages.

The disruption was not limited to chat. Voice mode, image generation, and Codex — OpenAI’s coding assistant — all stopped working simultaneously. Downdetector reports surged past 13,000 at peak, with the United Kingdom alone recording over 8,700 reports.

How OpenAI Responded

OpenAI initially labelled the issue as “degraded performance” on its status page before upgrading it to a “partial outage” as reports mounted. The company deployed a fix around 1 PM ET, and the service appeared fully recovered within 30 minutes.

OpenAI has not disclosed a root cause. The company said only that it was “monitoring the recovery.”

Why This Matters

ChatGPT now has hundreds of millions of active users, including businesses that rely on it for customer support, coding, content creation, and research. A 90-minute outage during working hours underscores how dependent many workflows have become on a single AI service.

Desktop web users were hit hardest, while some mobile app users retained partial access — a useful detail to remember if you experience future disruptions.

What This Means for You

If you rely on ChatGPT for daily work, today is a good reminder to have a backup plan. Tools like Claude, Gemini, and local models can serve as alternatives when any single service goes down. Consider keeping at least one secondary AI tool set up and ready.

For more on building resilient AI workflows, see our guide to choosing the right AI tool or visit our getting started page.