Claude is answering at its usual speed from all 10 countries we time it from.
Our probes' last round was at 14:07 UTC. Anthropic's status page lists Claude as degraded (read 14:16 UTC). Open the status page
- Some users unable to sign up with Apple on the Claude iOS app
Claude over the last 24 hours
Status pageClaude
Degraded
Our timingapi.anthropic.com
On time from 10 countries
Our timingclaude.ai
On time from 10 countries
The bands in words
- Anthropic's status page for Claude: Partial outage 13 Nov, 14:17 to 14:38 UTC; degraded 13 Nov, 14:38 to 21:50 UTC; degraded 02:56 to 04:23 UTC; degraded 07:14 to 14:17 UTC. Operational the rest of the time.
- Our timing of api.anthropic.com: Delayed 00:22 to 01:07 UTC; delayed 10:37 to 11:07 UTC. On time the rest of the time.
- Our timing of claude.ai: On time for the whole 24 hours.
What the status page listed
-
Incident Some users unable to sign up with Apple on the Claude iOS app
Claude: Investigating. Reported 07:11 UTC; still listed as open. Affects claude.ai.
We are investigating an issue with Sign in with Apple on the Claude iOS app, which is mainly affecting new accounts. New users can sign up with Google or email instead. We will provide an update as soon as possible.
The provider's newest update, 07:11 UTC
Meanwhile our probes confirmed: api.anthropic.com was delayed from Lauterbourg, France (10:37 to 10:52 UTC).
-
Incident Elevated purchasing errors on claude.ai
Claude: Resolved. Reported 02:54 UTC, resolved 04:23 UTC. Affects claude.ai.
The issue affecting purchases and plan changes on claude.ai has been resolved, and delayed plan changes and payments have now been applied. Impact occurred from 18:34 PT on November 13 / 02:34 UTC on November 14 to 20:03 PT on November 13 / 04:03 UTC on November 14.
The provider's newest update, 04:23 UTC
Meanwhile our probes confirmed: claude.ai was not answering from Sydney, Australia (03:52 to 04:07 UTC); claude.ai was not answering from Montreal, Canada (03:52 to 04:07 UTC); claude.ai was not answering from London, United Kingdom (03:52 to 04:07 UTC).
-
Incident Elevated errors affecting the Microsoft 365 connector
Claude: Resolved. Reported 12 Nov, 17:52 UTC, resolved 13 Nov, 21:49 UTC. Affects Claude Cowork and claude.ai.
This incident has been resolved. From 08:03 PT / 16:03 UTC on November 12 to 08:40 PT / 16:40 UTC on November 13, mail, calendar and search tools in the Microsoft 365 connector and Claude for Outlook returned errors for many requests because of a Microsoft 365 service incident.…
The provider's newest update, 13 Nov, 21:49 UTC
Our probes confirmed no delay on the way in to api.anthropic.com and claude.ai during this time.
-
Incident Elevated errors on claude.ai, the Claude API, and Claude Code
Claude: Resolved. Reported 13 Nov, 13:09 UTC, resolved 13 Nov, 15:38 UTC. Affects Claude Cowork, Claude API (api.anthropic.com), claude.ai and Claude Code.
The issue affecting multiple surfaces has been resolved.
The provider's newest update, 13 Nov, 15:38 UTC
Our probes confirmed no delay on the way in to api.anthropic.com and claude.ai during this time.
What our probes confirmed
- api.anthropic.com from Lauterbourg, France (Contabo network)
14 Nov, 10:37 to 10:52 UTC; up to 910 ms against a usual 168 ms, in 2 rounds. - claude.ai from Sydney, Australia (VERGETEL network)
14 Nov, 03:52 to 04:07 UTC; no answer in 2 rounds. - claude.ai from Montreal, Canada (OVH network)
14 Nov, 03:52 to 04:07 UTC; no answer in 2 rounds. - claude.ai from London, United Kingdom (OVH network)
14 Nov, 03:52 to 04:07 UTC; no answer in 2 rounds. - api.anthropic.com from Lauterbourg, France (Contabo network)
14 Nov, 00:22 to 00:52 UTC; up to 412 ms against a usual 168 ms, in 2 rounds.
- Operational, or on time
- Degraded, partial outage or maintenance; or delayed
- Major outage, or no answer
- No state given, or too few samples
- No report read, or no timings
- How we measure
Your browser sends three small requests to api.anthropic.com and claude.ai, with no cookies. The provider sees your IP address, as it does whenever you use Claude. The result stays on this page and is not sent to us.
api.anthropic.com
Claude's API and Claude Code send their requests to this address. Our request carries no key, so Anthropic's server refuses it at once. The time covers the network path and that refusal. No model runs.
| Timed from | Usual | Now | Last 24 hours | Reading |
|---|---|---|---|---|
| Your network | ||||
| United Statesyour comparison | 94 | 88 | On time | |
| Canada | 76 | 118 | On time | |
| Brazil | 156 | 160 | On time | |
| United Kingdom | 121 | 132 | On time | |
| Germany | 145 | 148 | On time | |
| France | 152 | 140 | On time | |
| India | 238 | 245 | On time | |
| Singapore | 252 | 277 | On time | |
| Japan | 184 | 196 | On time | |
| Australia | 232 | 244 | On time |
claude.ai
Cloudflare answers for claude.ai with a browser check before a request reaches Anthropic. This time shows how fast that front door answers, and nothing about Claude behind it.
| Timed from | Usual | Now | Last 24 hours | Reading |
|---|---|---|---|---|
| Your network | ||||
| United Statesyour comparison | 44 | 36 | On time | |
| Canada | 37 | 42 | On time | |
| Brazil | 18 | 16 | On time | |
| United Kingdom | 22 | 22 | On time | |
| Germany | 34 | 32 | On time | |
| France | 36 | 32 | On time | |
| India | 16 | 16 | On time | |
| Singapore | 20 | 45 | On time | |
| Japan | 18 | 22 | On time | |
| Australia | 18 | 18 | On time |
Why is Claude slow?
Right now the way in to Claude answers at its usual speed from every country we time it from. If Claude is slow for you, the time is going somewhere our probes do not reach. Slowness can sit in three places.
- Your own connection. Wi-Fi, mobile data, a VPN or a company proxy sit between you and Claude, and our probes do not share that path. The button above times the same hosts from your browser: if your times are slow while the probes in your country are not, the delay is on your side.
- The answer being written. Our times end at the first byte of the server's reply. A model writes after that, piece by piece, so long answers, hard prompts and modes that reason before answering take longer, and a long conversation gives it more to read before it starts. We time none of this.
- The provider. A busy or failing model is for Anthropic to report. Its status page is quoted at the top of this page and part by part below.
Claude Code
Claude Code sends its requests to api.anthropic.com, which this page times, as Anthropic's network configuration page lists. Set up to use Amazon Bedrock, Google Cloud or Microsoft Foundry instead, it sends them to that provider, and these boards do not time it.
Anthropic's status page gives Claude Code a line of its own: Operational
Reading the board
The probes are machines in datacenters, one or two per country. Every 15 minutes they time the same request, and the board shows the middle value. "Usual" is the middle value over the last seven days from the same probes.
The trace draws each round as a multiple of the usual time. The lower line is usual, the upper line is twice usual, and a yellow dot marks a round at or above it. One slow round is common and is not a delay. A country reads "Delayed" only when a probe is that slow in two of three rounds while it times another provider's hosts normally.
Anthropic's status page, part by part
- Degraded claude.ai
- Operational Claude Platform on AWS
- Operational Claude Console (platform.claude.com)
- Operational Claude in Microsoft Foundry
- Operational Claude API (api.anthropic.com)
- Operational Claude Models on Google Cloud (Anthropic managed)
- Operational Claude Code
- Operational Claude Cowork
- Operational Claude for Government
This is Anthropic's own statement, read from its public status page. Our timings are not part of it and do not change it.
Confirmed delays for Claude in the last seven days
- api.anthropic.com from Lauterbourg, France (Contabo network)
14 Nov, 10:37 to 10:52 UTC; up to 910 ms against a usual 168 ms, in 2 rounds. - claude.ai from Sydney, Australia (VERGETEL network)
14 Nov, 03:52 to 04:07 UTC; no answer in 2 rounds. - claude.ai from Montreal, Canada (OVH network)
14 Nov, 03:52 to 04:07 UTC; no answer in 2 rounds. - claude.ai from London, United Kingdom (OVH network)
14 Nov, 03:52 to 04:07 UTC; no answer in 2 rounds. - api.anthropic.com from Lauterbourg, France (Contabo network)
14 Nov, 00:22 to 00:52 UTC; up to 412 ms against a usual 168 ms, in 2 rounds.