Elevated errors on Claude Opus 5

Anthropic said the Claude Opus 5 error spike was resolved, with problems supposedly back to normal by 2:03 PST / 9:03 UTC. But in the comments, the real mood was less “crisis over” and more “here we go again.” One of the shortest reactions, “Déjà vu,” basically summed up the thread: users sounded tired, unimpressed, and very familiar with this kind of outage drama.

The spiciest frustration came from people saying the bigger issue isn’t just downtime, but trust. One commenter complained that as these chatbots are marketed as more powerful, they also seem to deliver more confident-sounding nonsense in everyday use. That turned a simple outage post into a wider mini-revolt about reliability: not just “is it up?” but “can I trust it when it is up?”

Then came the receipts. One user said they were still getting failures well after the incident was marked resolved, while the status page showed everything green — a classic internet trigger if there ever was one. That sparked the familiar tech-user side-eye: if people are still locked out, is it really fixed?

And because no online pile-on is complete without jokes, another commenter imagined Claude taking “an extra day off,” right after previous breaks, with AI rivals apparently also needing “hydration breaks.” Even the competition got dragged in, with one user asking whether OpenAI’s side of the fence is any better. In other words: outage story on paper, full-blown trust-and-rivalry soap opera in the comments.