Is Anthropic down right now?
Live Anthropic status pulled from status.anthropic.com. We track the Claude API (claude-3, claude-4 families including Opus, Sonnet, and Haiku), Claude.ai, Claude Code, and the Anthropic Console. Anthropic also publishes per-region status — useful when only us-east is degraded but eu-central is healthy.
Major outage: AWS
Anthropic
Degraded performanceClaude API, Claude.ai, Claude Code.
Status from status.anthropic.com · last fetched 16s ago
Developer Platforms
Upstream returned 404
Atlassian source hosting and Pipelines.
Container image registry and Docker Desktop.
Source hosting, Actions, Pages, Packages, Copilot.
macOS / Linux package manager (uses GitHub).
JavaScript registry, install API, web app.
Python Package Index — pip install backend.
AI / ML APIs
Some users unable to sign up with Apple on the Claude iOS app — started 1 day ago
Delayed Costs Data in the Compliance API — started 2 days ago
Upstream returned invalid JSON
Upstream returned invalid JSON
AI code editor — chat, tab, composer.
AI pair programmer — completions, chat, agents.
Gemini API, Vertex AI, AI Studio.
Cloud / Hosting
Service disruption: Increased Error Rates — started 61 days ago
Changes to Cloudflare Workers are not producing Event Subscription events — started 5 days ago
Upstream returned 404
Upstream returned invalid JSON
Droplets, App Platform, managed databases.
Global app servers, Postgres, volumes.
Compute, Storage, BigQuery, Cloud Run.
Azure VMs, App Service, AKS, Azure SQL.
Static + Jamstack hosting, Functions, Forms.
Web services, cron, Postgres, Redis.
Next.js hosting, Edge, Serverless functions.
Databases / Data
Upstream returned 404
Upstream returned invalid JSON
Upstream returned invalid JSON
Impaired Cluster operations in AWS me-central-1 and AWS me-south-1 — started 73 days ago
Serverless MySQL, Vitess-backed.
Postgres, auth, storage, realtime, edge fns.
Payments / Infra
SMS Delivery Receipt Failures from Twilio to Claro Costa Rica — started 3 days ago
SMS Delivery Receipt Failures from Twilio to Claro Costa Rica — started 3 days ago
Upstream returned invalid JSON
Upstream returned 404
Upstream returned 404
Developer-first transactional email API.
Auth / Identity
Communication
Monitoring / Dev Tools
Upstream returned 403
Upstream returned invalid JSON
Upstream returned invalid JSON
Collaborative design, FigJam, Dev Mode.
Feature flags, experiments, releases.
Error monitoring, performance, replays.
DevZone aggregates publicly available status data and refreshes every 30 seconds. Status indicators reflect each provider’s own machine-readable feed; for incident details and authoritative information, always consult the official status page linked on each card.
Anthropic 90-day incident history
Last 90 days
67.78% incident-free
Quick facts about Anthropic
- Anthropic status is published at status.anthropic.com (Statuspage.io).
- Claude API, Claude.ai, and Claude Code are reported as separate components.
- Different Claude model families and regions can degrade independently.
- Anthropic uses prompt caching to lower latency — caching itself can degrade independently of the model API.
Frequently asked questions
Why am I getting 529 errors from the Claude API?
A 529 is "Overloaded" — Anthropic is rate-limiting all clients during a capacity event. It’s not an outage in the traditional sense, but functionally similar. Implement exponential backoff (and consider a fallback to OpenAI or Gemini for critical paths). The dashboard above will show "degraded" or "partial outage" during 529 storms.
Is Claude Code affected when the Claude API is down?
Claude Code uses the Claude API under the hood, so yes — if the API is down, Claude Code completions and edits will fail. The Claude Code card on the Anthropic status page is reported separately, but is dependent on the API.
Does Anthropic have a status API?
Yes — https://status.anthropic.com/api/v2/summary.json. Same Statuspage schema as OpenAI.
How do I fall back from Claude to another provider during an outage?
Use a router like LiteLLM, OpenRouter, or Vercel AI Gateway. They unify the request schema so you can fail over to OpenAI or Gemini with a single config change.