Anthropic says Claude API and Claude Code are in partial outage for Opus and Sonnet errors

#ainews #anthropic #claude #outage

Anthropic says Claude API and Claude Code are in partial outage for Opus and Sonnet errors

Anthropic is investigating elevated error rates across several Claude models, including Opus 4.8, Opus 4.7, Opus 4.6, and Sonnet 4.6. This is worth treating as breaking news for builders because Anthropic's status API lists the incident impact as major and marks the Claude API and Claude Code as in partial outage.

What Anthropic has confirmed

Anthropic opened the incident at 00:37 UTC on June 22, 2026. The incident title is: “Elevated Error Rates for Opus 4.8, Opus 4.7, Opus 4.6, and Sonnet 4.6.”

The latest public update at publication time says Anthropic is continuing to investigate. The affected components listed by the status API are:

claude.ai — partial outage
Claude API (api.anthropic.com) — partial outage
Claude Code — partial outage
Claude Cowork — partial outage

Anthropic's overall status summary currently says “Minor Service Outage,” but the incident itself is marked with major impact. That distinction matters: the whole platform may not be down, but teams using the named Claude models should expect failures or degraded reliability until Anthropic posts a fix or resolution.

What builders should do now

If Claude is in a production path, assume requests may fail intermittently and make the failure mode explicit:

route critical workloads to a fallback model where you have one;
add retries with backoff rather than tight retry loops;
pause non-urgent batch jobs that depend on Opus 4.8, Opus 4.7, Opus 4.6, or Sonnet 4.6;
watch error budgets and queue depth for agent systems using Claude Code or the Claude API;
check whether customer-facing features need a degraded-mode message.

For coding-agent workflows, the practical risk is not just a failed prompt. Partial outages can leave long-running tasks half-finished, so teams should be careful with automated commits, deploy-adjacent agents, or scripts that assume Claude Code completed a full task.

Caveats

Anthropic has not yet published a root cause, expected recovery time, or exact error-rate percentage. This post is based on the official Claude status page and status API as of publication time. If Anthropic resolves the incident quickly, the main takeaway is still useful: production Claude integrations need fallbacks and clear degraded-mode handling.