Claude reliability hiccup

Claude went dark for many users this week, and Anthropic said the problem was an “elevated rate of errors” tied to Sonnet 4.6, the model that powers much of the Claude experience. On April 7 and April 8, users reported chats that appeared to think forever, login failures, and conversations that never returned an answer. (Anthropic Status, The Independent) Anthropic’s public status history shows two closely spaced incidents. One entry says Sonnet 4.6 had elevated errors on April 6, and another says Claude.ai had elevated errors on April 7 that affected desktop, mobile, login, voice mode, and completed chats. (Anthropic Status History, Anthropic Status) The company said the April 7 Claude.ai issue was fixed after it “applied a fix” and saw success rates return to normal. Status trackers that mirror Anthropic’s page also showed the service as operational again by April 8, even though users had reported fresh disruption within the prior 24 hours. (Anthropic Status History, (statusgator.com)) This was not an isolated blip in a quiet week. Anthropic’s status page also lists a March 31 to April 1 incident that caused elevated timeout rates for both Claude Opus 4.6 and Sonnet 4.6, which means the company has been dealing with repeated reliability problems across flagship models in a short span. (Anthropic Status) What users saw on the front end matched those backend notices. Reporting from The Independent said Claude often looked like it was processing a request but never produced a response, which is the kind of failure that feels less like a crash and more like a frozen cashier line where the register never finishes the transaction. (The Independent) The outage story lands at the same time Anthropic is tightening the faucet on usage. NBC News reported on April 3 that Anthropic had started imposing new usage limits as it struggled to keep up with surging demand for Claude, especially for Claude Code, its coding-focused product. (NBC News) That pairing matters because outages and limits can look different to users but stem from the same pressure. One is the system failing in real time; the other is the company deliberately rationing access so the system does not fail even more often. This relationship is an inference based on Anthropic’s outage notices and NBC News’ reporting on demand pressure. (Anthropic Status History, NBC News) NBC News said complaints began in late March, with Claude Code users saying they were hitting limits much faster than before on seemingly simple tasks. That suggests Anthropic is not only fighting headline-grabbing outages but also trying to stretch finite computing capacity across a fast-growing customer base. (NBC News) Other reporting around the same period points in the same direction. PCWorld reported that Anthropic confirmed it had been “adjusting” five-hour usage limits for Free, Pro, and Max users during peak hours, and The Independent separately reported that new restrictions followed a surge in demand. (pcworld.com), The Independent) Anthropic is hardly alone in wrestling with the economics of popular artificial intelligence products, but Claude’s case is unusually visible because the service is sold as a premium assistant for both everyday chat and professional coding work. When a general chatbot stumbles, it is annoying; when a coding assistant fails mid-workflow or burns through limits early, it can interrupt a workday and push users toward rivals. (NBC News, Anthropic Status History) The immediate episode appears to be over. But the bigger story is that Anthropic is trying to do two hard things at once in April 2026: keep Claude reliably online and decide who gets how much access when demand outruns supply. (Anthropic Status, NBC News)