{"slug": "claude-ai-suffers-widespread-outage-across-all-its-models-on-august-18", "title": "Claude AI Suffers Widespread Outage Across All Its Models on August 18", "summary": "Anthropic's Claude AI suffered a widespread outage on August 18, affecting multiple models including Claude Mythos 5, Claude Fable 5, Claude Opus 5, Claude Sonnet 5, and Claude Haiku 4.5, as well as claude.ai, the Claude API, Claude Code, and Claude Cowork. The incident began at 16:20 UTC and was still under investigation at 17:12 UTC, with no root cause or recovery timeline provided. This follows a series of disruptions earlier in the week, with 90-day uptime figures ranging from 99.35% for claude.ai to 99.47% for Claude Cowork.", "body_md": "*Claude stumbled again on August 18, and Anthropic's own status page made the problem plain: errors hit multiple models and several user-facing services at once.*\n\nIf you tried to use Claude on Tuesday and got errors back, you weren't alone. Anthropic's status page opened an incident at 16:20 UTC on August 18 for degraded performance affecting multiple models, then named Claude Mythos 5, Claude Fable 5, Claude Opus 5, Claude Sonnet 5, Claude Haiku 4.5 and other Claude models in the same update. That's broad. By 17:12 UTC, the company was still investigating elevated errors on Claude Opus 5.\n\nThe useful thing about a status page is that it strips away the drama. The less useful thing is that it usually strips away the cause as well. Anthropic said it was investigating and would provide another update as soon as possible, but it didn't identify a root cause or give users a recovery timeline. That gap matters when your morning depends on the tool answering.\n\nThis wasn't just a chatbot annoyance. The same status page showed degraded performance for claude.ai, the Claude API, Claude Code and Claude Cowork, while Claude Console and Claude for Government were listed as operational when checked Tuesday. If you're only asking Claude to summarize a memo, an outage is irritating. If you've built your coding work, internal tools or agent workflows around the API, it's a work stoppage.\n\n## The week already looked rough\n\nTuesday's incident landed after a crowded run on Anthropic's own incident log. On August 17, Anthropic reported degraded performance for Claude Opus 5 and Claude Sonnet 5. On August 16, it logged a service disruption affecting claude.ai, platform.claude.com, the Claude API, Claude Code and Claude Cowork, after first saying it was investigating authentication trouble for claude.ai, Claude Code and Claude Cowork at 21:58 UTC. That incident was marked resolved at 22:34 UTC.\n\n[Claude Code can now tap through your iOS app in real time and fix what it finds](https://startupfortune.com/claude-code-can-now-tap-through-your-ios-app-in-real-time-and-fix-what-it-finds/)\n\nAnthropic launched iOS Simulator integration for Claude Code Desktop on July 21, putting a live app-testing pane directly in the conversation. Claude can now install, run, and tap through an iOS app in real time, iterating on bugs without requiring screen recording or Accessibility permissions , a first among major AI coding tools. - [how to fix iOS app bugs](https://startupfortune.com/claude-code-can-now-tap-through-your-ios-app-in-real-time-and-fix-what-it-finds/) - [iOS simulator integration in Claude Code](https://startupfortune.com/claude-code-can-now-tap-through-your-ios-app-in-real-time-and-fix-what-it-finds/)\n\nGo back another day and the pattern keeps filling in. On August 15, Anthropic said an issue affecting Claude Fable 5 had been resolved, with impact running from 20:00 UTC on August 14 to 00:11 UTC on August 15. On August 14, the company logged separate disruptions affecting Claude Code on the web and related services, plus another incident involving the Claude API, Claude Code and Claude Cowork. August 13 brought elevated errors for Claude Mythos 5, Claude Fable 5 and Claude Sonnet 5. August 12 brought degraded performance for multiple models.\n\nThat is a pattern. It doesn't prove the same fault caused every interruption, and Anthropic hasn't said that it did. But for users, the internal difference between a model error, an authentication failure and a Claude Code disruption matters less than the external result: the tool isn't there when you reach for it.\n\nThe 90-day numbers make the issue look smaller than the week feels. Anthropic's status page listed claude.ai at 99.35% uptime, the Claude API at 99.46%, Claude Code at 99.38% and Claude Cowork at 99.47% when checked during the August 18 incident. Those figures aren't disasters. They also don't capture the frustration of repeated failures packed into a few working days.\n\n## Reliability is now part of the pitch\n\nAnthropic has pushed Claude Code hard toward developers, and that changes the standard. A consumer chatbot can get away with occasional downtime more easily than a work tool sitting inside pull requests, terminal sessions and automated pipelines. That is not a side use case. This is paid work.\n\nOpenAI and Google have had their own outages as ChatGPT and Gemini became daily tools for more users, so Anthropic isn't facing a problem unique to Claude. Look at the shape of this one, though. Multiple model names. Multiple surfaces. A run of incidents across six straight dated entries on Anthropic's own page. You don't need a grand theory to see why users will start asking harder questions about resilience.\n\nFrankly, the status-page language is too thin for a platform this central to people's work. Investigating is a fine first update. It isn't enough when the same page has been busy all week. Users need the service back, but they also need to know whether Anthropic has found the weak point before the next workday starts.\n\nAs of Tuesday afternoon, Anthropic's August 18 incident remained unresolved on its status page. The company had narrowed at least one update to Opus 5 errors, but the broader record still showed degraded performance across several Claude services. That's the open question for anyone depending on Claude this week: not whether one incident clears, but whether the pattern does.\n\n[Anthropic Says China Was Never Authorized to Use Claude Code Anyway](https://startupfortune.com/anthropic-says-china-was-never-authorized-to-use-claude-code-anyway/)\n\nChina's Ministry of Industry and Information Technology warned that Anthropic's Claude Code contains a hidden monitoring mechanism sending user data to remote servers. Anthropic confirmed the code exists but said the affected users in China were never authorized to use Claude Code in the first place. - [claude code china unauthorized access](https://startupfortune.com/anthropic-says-china-was-never-authorized-to-use-claude-code-anyway/) - [anthropic tracking mechanism developer tools](https://startupfortune.com/anthropic-says-china-was-never-authorized-to-use-claude-code-anyway/)\n\n**Also read:** [AI Is Creating More Jobs Than It Cuts in the UK, Lloyds Finds](https://startupfortune.com/ai-is-creating-more-jobs-than-it-cuts-in-the-uk-lloyds-finds/) • [This Week in AI: Robots Take a Punch, Assistants Get Long-Term Memory, and Models Keep Getting Cheaper](https://startupfortune.com/this-week-in-ai-robots-take-a-punch-assistants-get-long-term-memory/) • [Groq Raised $350 Million to Become Nvidia's Landlord Instead of Its Rival](https://startupfortune.com/groq-raised-350-million-to-become-nvidias-landlord-instead-of-its-rival/)", "url": "https://wpnews.pro/news/claude-ai-suffers-widespread-outage-across-all-its-models-on-august-18", "canonical_source": "https://startupfortune.com/claude-ai-suffers-widespread-outage-across-all-its-models-on-august-18/", "published_at": "2026-08-18 18:02:07+00:00", "updated_at": "2026-08-18 18:13:10.679771+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-products", "ai-infrastructure"], "entities": ["Anthropic", "Claude Mythos 5", "Claude Fable 5", "Claude Opus 5", "Claude Sonnet 5", "Claude Haiku 4.5", "Claude Code", "Claude Cowork"], "alternates": {"html": "https://wpnews.pro/news/claude-ai-suffers-widespread-outage-across-all-its-models-on-august-18", "markdown": "https://wpnews.pro/news/claude-ai-suffers-widespread-outage-across-all-its-models-on-august-18.md", "text": "https://wpnews.pro/news/claude-ai-suffers-widespread-outage-across-all-its-models-on-august-18.txt", "jsonld": "https://wpnews.pro/news/claude-ai-suffers-widespread-outage-across-all-its-models-on-august-18.jsonld"}}