Skip to main content
  • Pricing
News

Claude API, Claude Code and Cowork Outages: What Anthropic's Status Page Recorded

By Oakgen TeamPublished 3 min read

Share
Claude API, Claude Code and Cowork Outages: What Anthropic's Status Page Recorded

Anthropic's status page has recorded more than one incident that listed the Claude API, Claude Code and Claude Cowork together, so there is no single "the outage". This record covers four of them, read directly from status.claude.com on 2026-11-19, in UTC. None of the four entries states a root cause.

The four incidents at a glance

Date (2026)First post (UTC)Stated impact window (UTC)Marked resolved (UTC)Entry title
Sep 2914:2114:00 to 14:5916:27Elevated errors on claude.ai, Claude Code, Claude Cowork and the Claude API
Oct 2715:0914:42 to 15:1715:50Elevated errors for Claude Sonnet 5.5
Nov 1313:09Not stated15:38Elevated errors on claude.ai, the Claude API, and Claude Code
Nov 1909:1008:56 to 09:1109:33Elevated errors for Claude Mythos 5.2 and Claude Fable 5.2

All four entries end with a footer naming the affected components. For these four it lists claude.ai, the Claude API (api.anthropic.com), Claude Code and Claude Cowork. The September 29 entry also lists the Claude Console (platform.claude.com). The footer shows which components were attached to the incident. It does not say how badly each one was hit.

September 29: errors, then a separate sign-in problem

The first post at 14:21 UTC described elevated error rates on claude.ai, including the desktop and mobile apps, plus Claude Code and Cowork. It did not mention the API. The 14:28 UTC update added that some requests to the Claude API were also returning errors, and the 14:30 UTC update widened the list to claude.ai, platform.claude.com, the Claude API, Claude Code and Cowork. At 14:41 UTC Anthropic reported a mitigation and sharply lower error rates, with some sign-ins, account actions and Claude Code and Cowork sessions still failing. At 15:00 UTC it said the errors that began at 14:00 UTC had been mitigated at 14:36 UTC, but a second issue was blocking many sign-ins, including single sign-on and Sign in with Apple, along with new chats, voice conversations, Claude Code and Cowork sessions, purchases and file uploads. Signed-in users were asked not to sign out.

The 15:11 UTC update said most services had recovered as of 14:59 UTC and that some messages sent between 14:00 and 14:59 UTC may not have been saved. Only the 15:30 UTC update said all services had been operating normally since 14:59 UTC. The resolved note at 16:27 UTC gives the impact as 07:00 to 07:59 PT, or 14:00 to 14:59 UTC.

October 27: one model, four surfaces

This entry is short. It was posted at 15:09 UTC as an investigation into elevated errors on requests to Claude Sonnet 5.5, and the resolved note gives the impact as 07:42 to 08:17 PT, which is 14:42 to 15:17 UTC, a 35-minute window. Its title names a model rather than a product, yet the footer lists the API, Claude Code and Cowork alongside claude.ai.

November 13: the longest public update trail

The first post at 13:09 UTC mentioned degraded performance on claude.ai, the API and Claude Code. Later updates widened the list: connectors including Claude Docs and Claude Sheets at 13:52 UTC, then Claude Security and Claude Managed Agents at 14:03 UTC. At 14:36 UTC the page reported recovery on claude.ai and the API but warned errors might return. At 15:33 UTC it said errors on claude.ai, the mobile apps, the API and Claude Code were back to normal levels, existing connectors worked again, and Managed Agents and Meetings were recovering.

That update also listed leftovers: adding a new connector on claude.ai was unavailable, scheduled tasks and routines could run more than two hours late with some runs skipped, and Claude Security scans, reviews and project setup were delayed. The entry was marked resolved at 15:38 UTC. It gives no impact window, so the 2 hours 29 minutes between first post and resolution is the span of the public updates, not a measured outage length.

November 19: the shortest

The first post at 09:10 UTC covered elevated errors on requests to Claude Mythos 5.2 and Claude Fable 5.2. At 09:22 UTC the page said a fix had been deployed for other Claude models and it was monitoring recovery. The resolved note at 09:33 UTC gives impact from 00:56 PT / 08:56 UTC to 01:11 PT / 09:11 UTC, so the first public post came 14 minutes into that window, one minute before it ended.

Confirmed and not confirmed

Confirmed by the status entries: the timestamps above, the component lists, the symptoms described, the mitigation steps and the stated impact windows for three of the four.

Not stated in them: a root cause, the number of affected users or requests, whether API customers were credited, and how much data was lost on September 29 beyond the "may not have been saved" warning.

For context, the history view lists 13 incidents for September, 17 for October and 12 for November 2026 across all components, and the uptime view shows 99.69%, 99.85% and 98.89% for the claude.ai component in those months, with November still in progress when read. Those figures cover one component and say nothing about the API or Claude Code on their own.

Reading a status entry

Compare three things: the time of the first post, the stated impact window and the resolved time. They can differ by more than an hour, as on September 29. Most resolved notes give both Pacific and UTC times, so convert before comparing against your own logs.

Take this further with Oakgen Agent

Agent can research the public status page with web search and page fetch, write the runbook and checklist, and produce it as a downloadable docx or md file.

Coming soon: Continuous monitoring of a status page with alerts inside Agent is not available yet.

Open this task in Oakgen Agent

The prompt is filled in for you. Nothing is sent until you press send.

Open in Oakgen Agent

Frequently asked questions

Which Claude incident affected the API, Claude Code and Cowork together?

Several did. This article read four entries on status.claude.com in full where the incident footer lists claude.ai, the Claude API (api.anthropic.com), Claude Code and Claude Cowork: September 29, October 27, November 13 and November 19, 2026. The most recent began with a first post at 09:10 UTC on November 19, 2026 and was marked resolved at 09:33 UTC. The status page's recent list also shows resolved notices with the same wording on November 8 and November 16, which are not detailed here.

How long did the November 19, 2026 Claude outage last?

Anthropic's status entry states the impact ran from 00:56 PT / 08:56 UTC to 01:11 PT / 09:11 UTC, which is 15 minutes. The first public post came at 09:10 UTC, a fix for other Claude models was reported at 09:22 UTC and the incident was marked resolved at 09:33 UTC. The entry names elevated errors on requests to Claude Mythos 5.2 and Claude Fable 5.2.

Did Anthropic say what caused these outages?

Not in the four entries read for this article. They describe symptoms, affected components, mitigation steps and timings. They do not give a root cause, and this article found no separate post-incident write-up for them. Any cause you read elsewhere is not confirmed by these status entries.

Were messages lost during the September 29, 2026 incident?

The status entry says some messages sent between 14:00 and 14:59 UTC may not have been saved. The 15:11 UTC update said most services had recovered as of 14:59 UTC, with signing in, new chats, voice conversations, Claude Code and Cowork sessions, purchases and file uploads working again; the 15:30 UTC update said all services had been operating normally since 14:59 UTC. It does not say how many messages were affected.

Where can I follow Claude incidents as they happen?

Anthropic's status page at status.claude.com lists each incident with timestamped updates. As of 2026-11-19 the page offers subscriptions by email, SMS, Slack, Microsoft Teams, webhook and RSS or Atom feed, plus separate Incidents and Uptime views.

Sources

  1. Anthropic, Claude Status: Incident History, September to November 2026 (read 2026-11-19)
  2. Anthropic, Claude Status: Elevated errors for Claude Mythos 5.2 and Claude Fable 5.2, Nov 19 2026 (read 2026-11-19)
  3. Anthropic, Claude Status: Elevated errors on claude.ai, the Claude API, and Claude Code, Nov 13 2026 (read 2026-11-19)
  4. Anthropic, Claude Status: Elevated errors for Claude Sonnet 5.5, Oct 27 2026 (read 2026-11-19)
  5. Anthropic, Claude Status: Elevated errors on claude.ai, Claude Code, Claude Cowork and the Claude API, Sep 29 2026 (read 2026-11-19)
  6. Anthropic, Claude Status: Uptime History (read 2026-11-19)
Share