Claude: Elevated errors across all models – Resolved
Points and comments are a snapshot, not live.
Anthropic resolved elevated error rates across all Claude models after a multi-hour outage.
Claude experienced elevated error rates and increased latency across all models from approximately 12:45 PT on July 29, 2026. The issue was identified around 20:33 UTC, with recovery seen across most models by 21:38 UTC. Full recovery was confirmed by 22:36 UTC, and the incident was marked as resolved. The outage lasted roughly three hours, from initial investigation to resolution.
What commenters are saying
Commenters expressed frustration over downtime, with many noting reliance on AI coding tools. Some joked about forgetting how to code without assistance. Several users advocated diversifying across multiple AI subscriptions (e.g., Claude and Codex) to mitigate outages. Others discussed the trade-off between productivity and the loss of intrinsic coding skills. A few commenters reported specific errors like HTTP 522 and 500. The thread also debated Anthropic's capacity management versus OpenAI's approach, with some praising Claude's model quality while criticizing its uptime.