AI Models & Platforms

Claude Outage Hits Claude.ai, API, Claude Code and Cowork as Errors Spread Across Models

mm
Published
Add Unite.AI to your preferred sources on Google

Anthropic opened an incident on August 20, 2026, after elevated errors began hitting requests to what the company described as some Claude models, with the disruption reaching across nearly the entire Claude product line. The incident notice, posted at 19:16 UTC, lists claude.ai, the Claude API at api.anthropic.com, Claude Code, and Claude Cowork as affected.

The Claude status page shows all four components in partial outage, while Claude Console at platform.claude.com and Claude for Government remain operational. That split suggests the fault sits in the model-serving path rather than in account or administrative infrastructure: the consumer app, the developer API, and the two agentic products all draw on the same model backends, and all four are degraded together.

Anthropic’s initial notice is short even by status-page standards. The company said it is investigating elevated errors on requests to some Claude models and promised additional updates as soon as possible. As of the incident’s opening it had not named the affected models, attributed a cause, or given an estimated recovery time.

Third Incident on Anthropic’s Status Page in 24 Hours

The August 20 incident is Anthropic’s third service disruption in roughly 24 hours and the latest entry in a run of elevated-error events that has now stretched across eight consecutive days.

Hours earlier, at 18:32 UTC on August 20, Anthropic opened a separate incident for elevated errors on Google connectors, covering Sheets, Docs, Slides, and Chat integrations on claude.ai, which it marked resolved at 19:01 UTC, fifteen minutes before the multi-model incident was posted. On August 19, 2026, Anthropic identified the cause of elevated errors on Claude Opus 5 and Claude Haiku 4.5 at 09:42 UTC, implemented a fix and entered monitoring at 10:33 UTC, and resolved the incident at 11:02 UTC.

The pattern extends further back. On August 18, 2026, elevated errors hit Claude Mythos 5, Fable 5, Opus 5, Sonnet 5, Haiku 4.5, and other Claude models before narrowing to Opus 5, with recorded impact from 16:11 to 18:23 UTC. On August 17, 2026, Opus 5 and Sonnet 5 degraded in sequence. On August 16, 2026, an authentication issue disrupted claude.ai, the Claude Console, the API, Claude Code, and Cowork — an outage Unite.AI covered at the time — and on August 14 and 15, 2026, the company worked through separate disruptions affecting the API, Claude Code, Cowork, and Fable 5, the Fable 5 impact running from 20:00 UTC on August 14 to 00:11 UTC on August 15. Anthropic has not publicly linked any of these events to a common cause.

What the Outage Reaches

The affected component list matters because it spans every surface a production workload can touch. Claude Code and Claude Cowork, which Anthropic has named as flagship products, both run long, multi-step sessions against the API, so an elevated error rate interrupts agent runs mid-task, which is a different failure mode than a failed chat message. For teams that have wired Claude into CI pipelines, support tooling, or internal agents, the API partial outage is the load-bearing line on the page.

The resilience question this raises is the one the tooling market has been building toward: multi-model routing products such as Ramp’s Router.com gateway exist precisely because single-provider dependency turns a vendor’s bad week into the customer’s outage. Anthropic’s scale makes the dependency question sharper — Bloomberg reported a run-rate revenue of $65 billion earlier this month, a figure Reuters separately attributed to a person familiar with the matter, a figure that measures how much production traffic now rides on this serving stack.

The Numbers on the Status Page

  • 4 components in partial outage: claude.ai, Claude API (api.anthropic.com), Claude Code, and Claude Cowork
  • 2 components operational: Claude Console (platform.claude.com) and Claude for Government
  • 3 incidents in roughly 24 hours: Google connectors (18:32 UTC August 20, resolved 19:01 UTC), Opus 5 and Haiku 4.5 (August 19), and the current multi-model event
  • 8 consecutive days with at least one incident on the status page, from August 13 through August 20, 2026
  • 90-day uptime figures shown on the page: claude.ai 99.38%, Claude API 99.48%, Claude Code 99.4%, Claude Cowork 99.49%, Claude Console 99.86%, Claude for Government 100%

Anthropic said it would provide additional updates on the incident as soon as possible. The status page lists email, SMS, Slack, Microsoft Teams, and webhook subscriptions for incident notifications.

mm

Miles Okada is an AI-generated analyst at Unite.AI, covering artificial intelligence and cybersecurity with a focus on emerging threats, defensive architectures, and the evolving dynamics between attackers and automated systems. His work examines how AI is reshaping security operations, from autonomous threat detection and response to the rise of adversarial AI techniques.

With a technical and investigative perspective, Miles analyzes security research, incident disclosures, and real-world deployments to understand where AI strengthens defenses—and where it introduces new vulnerabilities. He pays particular attention to model exploitation, data poisoning, attack automation, and the operational realities of securing AI-powered systems at scale.

Articles authored by Miles Okada are AI-generated and reviewed by Unite.AI’s editorial team to ensure accuracy, rigor, and responsible coverage of the rapidly changing AI security landscape.