Open Claude on a weekday in August 2026 and there’s a fair chance you’ll meet a spinner, an error, or the words “server is busy” where an answer should be. It might be Claude Code stalling mid-task, the app refusing to load a chat, or the API throwing elevated error rates at whatever you built on top of it. You are not imagining it, and you don’t have to take our word for it: the receipts are on Anthropic’s own status page.
Answer first, because the shape of this is simple and the excuses are not. Claude has had a genuinely rough month. By Anthropic’s own count on status.claude.com, there were roughly twenty separate incidents in the first twenty-four days of August — degraded performance, elevated errors, and a couple of outright service disruptions — and as we hit publish on the 24th, a fresh one marked “major” was still unresolved, taking claude.ai, the API, Claude Code and Cowork down together. That is close to an incident a day. The reason it matters isn’t that software sometimes breaks; everything breaks. It’s who pays for it, and how.
We should say up front that Anthropic does a lot right here, and we’ll give it that in full below. A public, granular status page is more honesty than much of the industry offers, and most of these incidents were short. But “short and frequent” is still a pattern, and when you’re paying for a tool with a metered allowance, the frequency is the problem.
