Claude Had Error Problems on September 22: Here's What Happened
Anthropic reported a problem titled "Elevated errors for multiple models" on September 22, 2026. At 00:57 UTC, the company said it was looking into more failed requests than normal on Claude Mythos 5.1, Claude Fable 5.1, and Claude Opus 5. Anthropic Status
At 01:17 UTC, Anthropic said it had found the cause and was working on a fix. Those first updates gave no numbers, error codes, or regional details.
The report names four affected places: claude.ai, the Claude API (api.anthropic.com), Claude Code, and Claude Cowork.
Anthropic's Claude Status history lists 55 incidents. A separate entry titled 'Elevated errors across many models' is marked as resolved.
MacRumors reported an outage affecting multiple Claude models on July 6, 2026, resolved as of 09:13 UTC. MacRumors TechCrunch reported on March 2, 2026 that thousands of users reported problems reaching Claude services. TechCrunch
On September 3, 2026, The Verge noted in an update that outages affecting ChatGPT, Claude, and Grok had been resolved. The Verge TechCrunch reported on September 10, 2025 an outage that hit APIs, Console, and Claude, and on June 4, 2024 that Claude and Perplexity began seeing issues on Tuesday morning. TechCrunch
The broader context here is practical. Think of one kitchen serving dine-in, takeout, and delivery. If the kitchen has trouble, all three slow down. The same is true for direct use on claude.ai, apps using the API to ask Claude for answers, and helper tools in Code and Cowork. Builders cannot count on one path staying fine.
In practical terms for teams, more errors mean automatic retries, requests piling up, slower answers, and choices about a backup. It helps to write down which model was used, stay on a set version where allowed, and use the status page to pause less urgent work. When all three models are named together, switching names is unlikely to help. Reusing saved answers where okay and showing a clear message matter more.
In my view, the good part is fast communication. Anthropic reported the issue at 00:57 UTC and found the cause by 01:17 UTC. Online AI tools will have bad hours. Trust comes from this habit: name the models, name the places affected, update quickly, then share a fix. My kids grew up assuming the internet just works. Work in incident threads like this is what makes that expectation fair.
