Anthropic reports the Claude Code CLI operational. A one-off failure is most likely local — retry, then check the cases below.
A fix has been implemented and we are monitoring the results.
View Incident Detailarrow_forwardLearn more here: https://anthropic.com/news/fable-mythos-access
View Incident Detailarrow_forwardClaude Code
CLI Tool
Operational
Most "Claude Code is down" reports fall into a few buckets. Here is which are Anthropic's side and which are local.
What it means
The CLI talks to the same inference backend as everything else. A 529 here means Anthropic is at capacity, not that your install is broken.
What to do
Retry after a short wait, or switch the active model to a lighter one. During a real spike, give it a few minutes.
What it means
OAuth token expired, a stale credential cache, or an API key without access to the model you selected. This is usually local, not an outage.
What to do
Re-run the login flow, or check your API key and its model permissions. If many users report login failures at once, check the banner above.
What it means
Either backend degradation (component amber/red above) or a very large context/tool loop on your side eating time.
What to do
Check the live status above first. If it is green, try a smaller request or a fresh session; long agentic loops can stall on a single huge step.
What it means
You hit your usage or token limit — your account, not Anthropic. Returns HTTP 429.
What to do
Back off and retry, lower concurrency, or check your plan limits. This will not clear by waiting for an "outage" to end.
Read the banner at the top of this page. It is driven by Anthropic's own Claude Code component on the public status page: red means a confirmed outage, amber means degraded performance, green means operational. If it is green but you still see failures, the problem is most likely your install, network, or API key.
The Claude Code CLI uses the same inference backend as claude.ai and the API. An 'Overloaded' / HTTP 529 means Anthropic's servers are temporarily at capacity — not a bug in your setup. Retry with a short backoff, or switch to a lighter model.
If the banner and the Claude Code component card above are green, suspect your side first: expired login, wrong API key, model permissions, or a huge context. If the status above is red or amber, it is Anthropic's side and waiting is the fix.
Yes. Anthropic exposes 'Claude Code' as its own component on the public Statuspage, so the status and 30-day uptime bar above reflect the CLI tool itself, separately from claude.ai and the API.