ChatGPT, Codex, Claude and Grok all falter in one afternoon

Anthropic opened an incident titled “Elevated errors for multiple models” at 13:26 UTC (15:26 Brussels). By 15:25 UTC it said only Opus 4.8 and Opus 5 were still affected, and at 16:06 UTC that a fix had been deployed for Claude Mythos 5.1, Claude Fable 5.1 and Claude Opus 5 while it monitored recovery. OpenAI opened “Elevated errors across ChatGPT and Codex” at 14:58 UTC (16:58 Brussels), applied a mitigation and moved to monitoring at 15:50 UTC. In the same window a Codex issue on GitHub and an “Ask HN: why are OpenAI, Claude and Grok simultaneously down?” thread filled up, and xAI's status page reported problems with Grok. MacRumors wrote at 8:56 am PDT that ChatGPT, Claude and Grok were down or degraded for many users on iPhone and the web, adding in an update that the chatbots were coming back online. None of the three vendors has published a root cause, and no shared dependency has been confirmed — for coding agents the practical lesson is that a second provider only helps if the fallback is wired up before an afternoon like this. Ars Technica adds that DownDetector/StatusGator also showed a likely Gemini API blip (~10:45–11:15 ET) while AWS/Azure/Cloudflare had no major declared incidents — making four frontier stacks overlapping the same morning unusually rare.