Anthropic confirmed a Claude outage on September 3, 2026, after users began reporting elevated errors when sending requests to several of the company's AI models. The incident started around 9:41 AM ET, according to Anthropic's status page, and disrupted access to the company's newest Mythos, Fable, and Opus models.
The outage matters because Claude powers coding tools, customer support systems, and business workflows through Anthropic's API, so failed requests translate directly into stalled work for developers and companies that depend on it.
Key takeaways
- Anthropic confirmed elevated errors across multiple Claude models starting around 9:41 AM ET on September 3, 2026.
- Affected models include Mythos/Fable 5.1, Mythos/Fable 5, Opus 5, Opus 4.8, and Opus 4.6.
- Anthropic said it identified the cause of the errors and was working on a fix as of the initial status update.
- As of 10:49 AM ET, Anthropic said it was still working on a fix and the outage remained ongoing.
- Users see failed requests or error messages when using the affected models during the outage.
Affected
What happened to Claude on September 3?
The Claude outage is an incident in which Anthropic's AI models returned elevated errors instead of completing user requests, starting around 9:41 AM ET on September 3, 2026. Anthropic's status page first flagged the problem as affecting Claude Mythos 5.1, Claude Fable 5.1, and Claude Opus 5, before the company said it had identified the root cause and was preparing a fix.
Anthropic later widened the confirmed list of affected models to include Mythos/Fable 5.1, Mythos/Fable 5, Opus 5, Opus 4.8, and Opus 4.6. By 10:49 AM ET, the company said work on a fix was continuing, indicating the incident had not been resolved roughly an hour after it began.
- Outage began around 9:41 AM ET on September 3, 2026.
- Seven Claude model variants were confirmed affected.
Which Claude models are down?
Anyone sending requests to the affected Claude models, whether through the API, Claude apps, or integrated third-party tools, can encounter failed requests during the outage. Anthropic's confirmed list covers seven model versions across three product lines: Mythos, Fable, and Opus.
- Claude Mythos 5.1
- Claude Fable 5.1
- Claude Mythos 5
- Claude Fable 5
- Claude Opus 5
- Claude Opus 4.8
- Claude Opus 4.6
Why the Anthropic outage matters for businesses
Claude is embedded in developer tools, coding assistants, and customer-facing products through Anthropic's API, so an outage on this scale disrupts any workflow built on top of the affected models. Teams that route production traffic to Opus or Mythos/Fable without a fallback provider face stalled pipelines until Anthropic restores service.
Because the incident spans multiple current-generation models rather than a single deprecated version, it affects a broad slice of Anthropic's paying customer base at once, according to the company's own status page updates.
What does the error look like for users?
During the outage, users attempting to send prompts to an affected model see requests that either fail to complete or return an error response instead of a normal reply. Anthropic has not published a root-cause explanation beyond stating that the cause of the elevated errors was identified internally.
How is Anthropic responding?
Anthropic has acknowledged the incident through its public status page and provided incremental updates as the situation developed. The company said at 9:41 AM ET that it had identified the cause of the elevated errors, and by 10:49 AM ET confirmed it was still working on a fix.
Status page is the source of truth
Anthropic's status page remains the most current source for confirming whether service has been restored for a specific Claude model.
What should IT teams watch next?
Admins and developers relying on the affected Claude models should monitor Anthropic's status page for confirmation that a fix has been deployed and errors have subsided. Anthropic has not yet stated a resolution time, so teams running production workloads on Opus, Mythos, or Fable models should expect intermittent failures until an all-clear is posted.
Timeline
Impact & actions
The outage caused failed or errored requests for anyone using the affected Claude models via API, web, or app, disrupting workflows built around Claude during the incident window.
Recommended actions · Medium urgency
- 1Monitor Anthropic's status page for updates on the fix
- 2Add retry logic or fallback handling in applications calling the affected Claude models
Response
Vendor
Customer guidance
Check Anthropic's status page for live updates and expect intermittent failures on the affected models until the fix is deployed.
Updates
Anthropic said it identified the cause of the elevated errors and was working on a fix.
Anthropic confirmed it was continuing to work on a fix and the outage remained ongoing.
FAQ
Which Claude models are affected by the September 3 outage?
Anthropic confirmed elevated errors on Claude Mythos 5.1, Fable 5.1, Mythos 5, Fable 5, Opus 5, Opus 4.8, and Opus 4.6.
When did the Claude outage start?
The outage began around 9:41 AM ET on September 3, 2026, according to Anthropic's status page.
Has Anthropic fixed the outage?
As of Anthropic's 10:49 AM ET update, the company said it was still working on a fix and the outage remained ongoing.
What does the outage look like for users?
Users of the affected models see requests fail to complete or return error messages when using Claude via the API, web, or apps.
Is there a workaround for the Claude outage?
No workaround has been confirmed; Anthropic has not published a patch or mitigation beyond its ongoing fix.
The bottom line
Anthropic confirmed an outage affecting seven Claude models on September 3, 2026, with elevated errors first reported around 9:41 AM ET and no full resolution as of the last update.
What happens next
What to do
