Is Claude AI Down? Global Anthropic Outage Disrupts Chat, API and Claude Code Services

Anthropic's Claude AI experienced a global service disruption Wednesday affecting several flagship models, the Claude API and Claude Code, the company said. "We have identified the cause of elevated errors on requests to Claude Mythos 5, Claude Fable 5, Claude Opus 5, and Claude Sonnet 5 and are working on a fix," Anthropic said in its first incident update posted on the company's public status page at 07:05 UTC. The incident remains unresolved as of the latest update. While some Anthropic services continue operating normally, users worldwide are reporting failed chat requests, interrupted coding sessions and API errors that have disrupted development and enterprise workflows. Anthropic confirmed the outage through its public status page rather than social media. In its 09:13 UTC status update the company said engineers continue to work on a fix without providing an estimated time for service restoration or additional technical details. The degraded performance specifically affects Claude Mythos 5, Claude Fable 5, Claude Opus 5 and Claude Sonnet 5 with elevated error rates impacting chat generation, coding assistance and API requests. Anthropic's status page listed Claude Console and Claude for Government as operational throughout the incident while marking the Claude web app, API, Claude Code and Claude Cowork as experiencing degraded performance. Downdetector and community discussions across X indicate the disruption is affecting users across multiple regions. The most common problems reported include API Error 529, failed chat requests, Claude Code interruptions, request timeout errors and intermittent message failures. Wednesday's incident is one of several service disruptions Anthropic has experienced during 2026. The company has logged multiple elevated-error events throughout June, July and early August including a major outage in late June that generated thousands of user reports. Most previous incidents were resolved within one to two hours after engineers identified the problem. The August 5 disruption has already lasted several hours making it longer than many recent degraded-performance events.