Anthropic's Claude chatbot has experienced a series of outages over several days, with the AI company battling repeated bouts of elevated errors linked to Sonnet 4.6, the underlying model that powers Claude and several other parts of its product suite.
The disruptions, which affected users globally, have prompted thousands of reports on Downdetector, the outage tracking platform, and drawn attention to questions about the stability of Anthropic's infrastructure at a time of rapidly growing demand.
The first significant incident struck on 6 April, with Downdetector registering approximately 2,700 user reports within minutes as Claude users found themselves unable to log in, unable to use voice mode, and unable to complete standard chat interactions.
Anthropic updated its status page promptly, confirming elevated errors on Claude.ai across desktop and mobile, and said it was working to resolve the issue.
A fix was implemented by early afternoon, with the company shifting to monitoring status and reporting that success rates were recovering across affected services.
The reprieve proved short-lived.
A further incident on 7 April produced similar symptoms, with Downdetector reports peaking at more than 3,000 around midday Eastern Time as users again reported the system appearing to think indefinitely without returning any response.
Anthropic confirmed that incident too, acknowledging elevated errors affecting login, voice mode and chat before applying another fix within roughly two hours.
The following morning, on 8 April, a third and more sustained outage struck, with Sonnet 4.6 exhibiting errors above baseline from around 23:00 Pacific Time the previous evening until approximately 01:50 Pacific Time, before a second wave of issues began at 06:23 and was not resolved until 09:33, representing a disruption window of more than three hours.
The affected services included Claude.ai, the Claude API, Claude Code, Claude Cowork and the platform console, meaning the outages touched both consumer users and developers building on Anthropic's systems.
In total, the affected period between 6 and 8 April encompassed at least five separate declared incidents, ranging from brief warning-level disruptions to major outages lasting over an hour.
The pattern of failures comes as Anthropic is in a period of aggressive expansion, with its user base understood to have grown substantially following the launch of Claude Code and deeper enterprise integrations.
Analysts and observers noted that while Claude's incidents were resolved more quickly than a comparable 12-hour blackout suffered by OpenAI's ChatGPT in January 2026, the consecutive nature of the disruptions was likely to test user confidence among developers and businesses that depend on the platform for production workloads.
Related reading
- Anthropic spots 'emotion vectors' inside Claude
- Anthropic cuts off Claude Code subscribers from third-party AI agents, starting with OpenClaw
- Memento-Skills lets AI agents rewrite their own capabilities without retraining underlying models
Anthropic said the incidents had been resolved and confirmed it was continuing to monitor performance closely.
The company has not publicly detailed the underlying cause of the repeated failures or outlined what steps it is taking to prevent further recurrence.The recap
- Anthropic's Claude chatbot experiences outage causing user errors
- Users report systems get stuck thinking without delivering responses
- Company said Sonnet 4.6 showed an "elevated rate of errors"