Elevated errors across many models
claude.ai, Claude API (api.anthropic.com), Claude Code, Claude Cowork
Claude API Status
Check whether Claude API requests, Anthropic API 429 or 529 errors, timeouts, 500/503 responses, model latency, or Claude Code failures match an official Claude incident.
Current answerClaude is currently operational according to the official Claude status page.
Likely impactAPI requests, model responses, timeouts, 5xx errors, rate limits, and Claude Code dependencies
Next checkCompare the API component with your status code, model, key, quota, and request logs.
Fast answer
Claude API 429 vs 529 vs timeout vs 500/503
Start by classifying the error before changing code. The fastest fix is different for account rate limits, temporary overload, long-running requests, and service-side 5xx failures.
429 rate limit
Usually account, workspace, quota, burst traffic, or retry pressure.
529 overloaded
Usually temporary Anthropic-side overload or upstream capacity pressure.
Timeout
Often long prompts, streaming gaps, network route, model latency, or slow tools.
500 or 503
Can be transient service errors, degraded models, maintenance, or incident windows.
Model unavailable
Fable, Mythos, Opus, or auto mode errors can be model-specific rather than a full API outage.
Anthropic API 429, 529, timeout, 500, or 503
Start with the HTTP status code, then compare it with the live Claude API component and recent incidents. The same symptom can mean different things depending on whether it is account-specific, model-specific, or affecting many requests at once.
- 429 usually points to rate limits, quota pressure, bursty traffic, workspace limits, or retry storms.
- 529 overloaded usually points to temporary Anthropic-side or upstream capacity pressure.
- 500 or 502 can indicate transient service errors, model instability, or a recent upstream incident.
- 503 often means temporary unavailability, overload, maintenance, or a service-side degradation window.
- Timeouts require checking prompt size, streaming behavior, model latency, network route, and retry policy.
- If the same failures also appear in Claude Code, check whether Claude Code is depending on the same API component or model.
Claude API Component
The Claude API component covers developer requests to api.anthropic.com, model responses, elevated errors, and API-side degradation.
| Component | Status | What it affects |
|---|---|---|
| Claude API (api.anthropic.com) | operational | API requests to api.anthropic.com. |
Anthropic API error checklist
Use this checklist when you see Anthropic API error responses, failed Claude API requests, elevated latency, or repeated timeouts in production logs.
- 429: check rate limits, quota, workspace limits, request bursts, and retry behavior.
- 529: check official incidents, retry with backoff, reduce request size, and avoid retry storms.
- 500, 502, or 503: compare your logs with active and recent Claude API incidents.
- Timeouts: compare model choice, prompt size, streaming behavior, network route, and API latency.
- Authentication errors: verify API key, workspace, billing state, and model access.
- Model-specific failures: test whether the same request works on another available Claude model.
- Model temporarily unavailable: check Claude Code auto mode, explicit model selection, recent release notes, and whether the same request works outside Claude Code.
- Request-log clues: record status code, request ID, model, timestamp, latency, retry count, and prompt size.
- Claude Code failures: compare Claude Code status after checking the API component.
Recent Claude API Incidents
These official incidents affected the Claude API directly or appeared alongside broader Claude model and Anthropic service issues.
resolved
Elevated errors on Claude Fable 5 and Claude Mythos 5
claude.ai, Claude Console (platform.claude.com), Claude API (api.anthropic.com), Claude Code, Claude Cowork
Elevated errors on Claude Opus 4.8
claude.ai, Claude API (api.anthropic.com), Claude Code, Claude Cowork
We’ve suspended access to Claude Mythos 5 and Claude Fable 5
claude.ai, Claude API (api.anthropic.com), Claude Code, Claude Cowork
Elevated error rates on Opus 4.8
claude.ai, Claude Console (platform.claude.com), Claude API (api.anthropic.com), Claude Code, Claude Cowork, Claude for Government
Elevated errors on Opus 4.5
Claude API (api.anthropic.com)
Elevated error rate on Claude Opus 4.8
claude.ai, Claude Console (platform.claude.com), Claude API (api.anthropic.com), Claude Code, Claude Cowork
Claude API fails, but the web app may still work
API failures can come from service incidents, model availability, local request configuration, account limits, network routing, or authentication state.
- Check whether the Claude API component has an active or recent official incident.
- Compare your HTTP status code with the recent incident descriptions.
- Separate 429 rate limits from 5xx service errors before changing retry logic.
- For 529 overloaded, use the focused 529 page to check retry pressure, request size, and service overload.
- For request timed out or stream idle timeout errors, use the timeout page to separate service latency from long context, network route, and Claude Code workflow issues.
- Confirm the model name, API key, workspace, quota, and rate limit state.
- If requests are slow but not failing, compare the Claude slow page with your request logs.
- If Claude Code also fails, check Claude Code status after checking the API component.
- Avoid changing production retry or fallback logic until you know whether the issue is service-side or account-side.
FAQ
Why am I seeing an Anthropic API error?
Anthropic API errors can come from a service incident, elevated 5xx errors, timeouts, rate limits, invalid API keys, model availability, quota, network routing, or request configuration.
Where can I check Claude API status?
Use this page to check the official Claude API component, recent Anthropic incidents, active outages, and whether your API symptoms match a service-side issue.
Is Claude API down right now?
The current answer is shown at the top of this page. If the Claude API component or a related incident is degraded, API requests may fail, time out, or return elevated errors.
Can Claude API be down while Claude.ai works?
Yes. Claude.ai, Claude API, Claude Code, and Console can have separate incidents, so the web app may work while developer API requests fail.
Are 429 errors a Claude API outage?
Usually no. 429 errors normally mean rate limits or quota pressure unless the official Claude API component also shows an incident or many users report elevated errors.
What does Claude API 529 overloaded mean?
529 overloaded usually means Anthropic or a related upstream service is temporarily overloaded. Treat it as a service-pressure signal first, then check official status, recent incidents, retries, and whether the same model fails for other users.
Are Claude API 500 or 503 errors an Anthropic outage?
Repeated 500, 502, or 503 responses can be service-side, especially when they match a recent Claude API incident. Confirm with the API component, recent incidents, and whether the same model fails across multiple requests.
Why is the Anthropic API timing out?
Timeouts can be caused by API-side latency, model congestion, long prompts, network routing, retries, or a recent incident. Compare status, request logs, model choice, and timeout settings.
How do I tell rate limits from a Claude API outage?
Rate limits usually show 429 responses, burst patterns, quota pressure, or account-specific failures. Outages usually affect many requests, users, models, or components and may appear in official incidents.
What request logs should I check for Anthropic API errors?
Check HTTP status code, request ID, model name, timestamp, prompt size, streaming mode, retry count, latency, workspace, and whether failures cluster around one model or region.
Does Claude Code depend on Claude API status?
Claude Code can be affected by API-side model availability, authentication, and upstream Anthropic service incidents, so check API status when Claude Code commands fail.
Why does Claude Code say claude-opus-4-7 is temporarily unavailable?
This usually points to model routing, auto mode, model availability, or a Claude Code/API-side availability issue rather than a local prompt problem. Check Claude API status, Claude Code status, recent release notes, model selection, and whether an explicit model works.
What should I check before changing production code?
Compare the official API component, recent incidents, response status codes, model name, request logs, API key, account limits, retry policy, and network route.