Anthropic’s Claude platform experienced elevated error rates on Thursday evening, May 14, the company’s status channel reported, leaving users with intermittent failures as engineers investigated the problem. The update, posted at 8:48 p.m. ET, named Opus 4.6 and 4.7 — two recent model variants in the Claude family — as showing “elevated error rates,” and said the team was continuing to probe the issue.

Claude Status’s brief bulletin did not provide a root-cause diagnosis or an estimated time to full restoration. “We are continuing to investigate this issue,” the company wrote, signaling that the problem had been detected and that mitigation and troubleshooting were underway. The post did not detail whether outages were limited to particular geographies, customers or API versus hosted-chat usage.

Opus 4.6 and 4.7 are among the model variants that power Claude’s responses across Anthropic’s products; elevated error rates for those variants can translate into degraded performance or failures for applications and services that route queries through them. The status message suggested the company was handling the incident at the model-infrastructure level rather than citing a wider networking or third-party cloud provider outage.

By contrast, other large conversational AI services such as OpenAI’s ChatGPT were reported to be functioning normally at the same time, providing an alternate option for users who rely on multiple models for redundancy. The timing follows a period of sporadic service interruptions across several major internet platforms in recent weeks, when social apps and team-messaging services briefly logged widespread complaints before recovering — a context that highlights how reliant many customers have become on continuous API availability.

Anthropic did not immediately release additional technical details, and there was no public timeline for a fix in the status update. Customers and developers typically monitor the company’s status page and developer channels for progress notes; organizations that rely on Claude for production tasks often implement fallback routing to alternate models to reduce business interruption during such incidents.

For now, the situation remains under investigation. Further updates are expected from Claude Status as engineers work to contain the issue and restore normal service levels for Opus 4.6 and 4.7.

Continue Reading

Nvidia gains from easing China export controls as Cerebras IPO sparks AI inference competition
Next Story

Nvidia gains from easing China export controls as Cerebras IPO sparks AI inference competition

Popular Categories


Search the website