A ChatGPT login outage on 11 December 2024 knocked out sign-ins, new account creation, and the loading of previous conversations for users across the US and Europe, with TechRadar reporting that issues first appeared at 6:17 PM ET. The timing matters, because the original outage notice placed the start closer to 8 PM ET, a 90-minute gap that suggests the problem was spreading quietly before OpenAI’s status page caught up with it.
For users trying to access chatgpt.com, the experience was characteristically unhelpful: the sidebar would hang on a loading animation, messages would fail to send, and the service returned errors citing “too many concurrent requests.” Whether you were starting a fresh chat or trying to pull up an old conversation, the result was the same.
How Far Did the ChatGPT Login Outage Spread?
The disruption was not limited to the consumer-facing chat interface. OpenAI’s coding platform, Codex, was caught in the same failure. More consequentially for developers, as many as 12 API endpoints were listed as affected on OpenAI’s status page, the kind of breadth that ripples through every downstream application and service built on top of the platform.
OpenAI acknowledged the problems publicly and moved quickly in at least one respect: within roughly 14 minutes of the outage beginning, the company had updated its status page to mark the incident as “identified.” That is the correct procedural move, and worth crediting. What it did not mean, of course, was that the issue was fixed, only that OpenAI knew what it was dealing with and said it was “working on implementing a mitigation.”
By 8:15 PM ET, the incident remained ongoing. Users attempting their own tests continued to hit the same walls.
Service Restored by 9:30 PM ET
The outage did not drag on overnight. TechRadar reported that ChatGPT was working again as of 9:30 PM ET on 11 December, suggesting a total disruption window of somewhere over three hours from the first confirmed reports. For casual users that is an inconvenience; for developers relying on the API or businesses with integrations built on top of OpenAI’s services, even a three-hour window is the sort of gap that gets noticed in post-mortems.
OpenAI has not, at the time of writing, published a detailed incident report explaining what caused the failure. The status page update pattern (identified, mitigating, resolved) tells you the shape of the response but nothing about the underlying cause. That is standard practice for most cloud providers, at least in the immediate aftermath, though a postmortem document would be welcome given how many services were swept up at once.
Timing and the ChatGPT Login Outage in Context
The ChatGPT login outage landed at a notably busy moment for OpenAI. TechRadar noted that Sora, OpenAI’s text-to-video generator, had officially launched earlier that same week. Any major platform launch tends to push usage upward (more attention, more curiosity, more sign-ups) and whether that contributed to the strain on the login and API infrastructure on 11 December is an open question the company has not yet answered.
What is clear is that the failure was broad. Consumer logins, new account creation, the API, and Codex all showed problems simultaneously, which points to something closer to a shared authentication or infrastructure layer than an isolated service fault. OpenAI’s own status page, to its credit, reflected the scope accurately in near-real time.
According to TechRadar, service was fully restored by 9:30 PM ET, meaning the window from first confirmed disruption to recovery ran just over three hours. OpenAI’s status page is the place to watch for any formal incident review if one is published.
