Claude Services Degraded Amid Ongoing System Disruptions

Claude.ai, its API, and Claude Code are currently experiencing a system disruption with elevated error rates. Anthropic has identified the issue and is implementing a fix, though core services remain in a degraded state. This incident follows a series of frequent outages and performance issues documented throughout early April 2026.

Key Points

• Major Claude services including the web interface, API, and CLI tools are currently experiencing elevated error rates and degraded performance.
• Anthropic has identified the cause of the April 15 disruption and is in the process of deploying a fix.
• Historical data reveals a high frequency of incidents in April 2026, specifically targeting authentication, workspace creation, and model reliability.
• Global latency currently averages 97ms, though regional performance varies drastically from 14ms in India to over 500ms in China.
• Specialized services like Claude for Government and Claude Cowork have maintained higher stability compared to the standard consumer and developer platforms.

Sentiment

The community sentiment is predominantly negative and frustrated. Most commenters are experiencing the outages firsthand and express frustration with paying for unreliable service. There is heavy sarcasm around Anthropic's claims about solving coding while failing to maintain basic service uptime. However, the criticism comes from a place of dependency rather than dismissal, as many users clearly still prefer Claude's capabilities and are reluctant to switch. Hacker News largely agrees that the reliability situation is unacceptable but remains split on whether this is a temporary growing pain or a fundamental infrastructure failure.

In Agreement

• The outages are real and recurring, happening daily around 14:30 UTC when US Pacific users come online, confirming a pattern of insufficient capacity during peak hours
• The uptime statistics are alarmingly low, with users reporting sub-90% availability over 30 days, far below enterprise standards
• The problems extend beyond inference GPU capacity into authentication and login systems, indicating broader infrastructure issues
• Anthropic's status page sometimes lags behind actual outages, with users seeing errors while the page still shows 'All Systems Operational'
• The frequency and severity of disruptions have been increasing throughout April 2026

Opposed

• Anthropic's status page is actually more transparent than competitors like AWS, Azure, and GCP, who often hide their downtime
• The outages are comparable to early Twitter's fail whale era or Reddit's frequent downtime, and those platforms survived and thrived despite reliability issues
• GPU supply constraints are a global issue that no amount of valuation money can immediately fix
• Claude Code's actual code quality remains superior to alternatives even when available, and short outages don't negate the productivity gains
• Some users don't mind the downtime as long as it resolves in an hour, since it replaces manual coding that also had its own 'downtime' from breaks and distractions