Elevated errors on Claude Opus 5

Anthropic says the Claude Opus 5 outage is now fixed, with errors back to normal after several hours of trouble that hit the main app, the developer service, Claude Code, and even Claude Cowork. But if the official update was calm and tidy, the community reaction was anything but. Users flooded the discussion with reports of broken requests, overloaded messages, and a lot of exasperated sighing from people whose workday got body-slammed by the downtime.

The loudest theme was simple: trust is taking a hit. One frustrated user joked that “elevated errors” could just as easily describe the model itself, saying Opus 5 has been unreliable for coding and keeps introducing fresh problems. Another went even harder, claiming the bot was not only making more mistakes but sounding strangely off, as if it had lost its usual clear voice. That turned a status-page blip into a bigger mini-scandal: was this just a temporary outage, or were people already unhappy with the product even when it was online?

Then came the survival strategies and side-eye. One commenter said these downtime bursts are exactly why they pay for multiple AI subscriptions—a very 2026 version of keeping backup snacks in the drawer. And in classic internet fashion, not everyone wanted the drama documented at all, with one blunt reply snapping, “stop posting these on hn.” So yes, the servers recovered, but the comments section made it clear the real issue is confidence: when people rely on these tools for daily work, even a short wobble can spark jokes, anger, and a full-on loyalty crisis.