Skip to content
keyforge

anthropic

chat
vision

Test your Anthropic API key.

Verify a Claude key, see which Opus / Sonnet / Haiku models it can call, and time the round trip — without burning tokens on a real message.

Get a keyAPI docs Pricing

Stateless proxy — keys never logged, stored, or persisted. What happens to your key →

Detected
Anthropic

What this key does

An Anthropic API key authenticates against the Messages API (Opus / Sonnet / Haiku across the Claude 4 line), the streaming endpoint, and the prompt caching feature. The key uses the x-api-key header (not Bearer) and the anthropic-version header is required. Calling /v1/models is the cheapest validation path.

How to get a Anthropic API key

  1. Sign in at console.anthropic.com.
  2. Open Settings → API Keys and click Create Key.
  3. Pick the workspace this key should belong to. Workspaces let you split production from dev.
  4. Copy the key (shown once) and paste it here to confirm it works.
  5. Set a per-workspace spend limit before issuing the key to anyone — including yourself.

Common errors and fixes

  • 401 Unauthorized: Key is invalid, revoked, or pasted with extra whitespace. Generate a new key from the provider console and try again.
  • 403 Forbidden: Key is valid but lacks permission for this resource. Check project / org / workspace scope, or that billing is set up for this key.
  • 429 Too Many Requests: You hit the per-minute or per-day rate limit. Wait a moment and retry, or upgrade your tier.
  • 404 Not Found: The endpoint or model id changed. Check the provider docs for the current path and model identifier.
  • 5xx: The provider is having issues. Check their status page before assuming the bug is yours.

Security best practices

  • Store keys in an env var or secret manager — never commit them to a repo, even a private one.
  • Restrict scope: prefer per-project or per-deployment keys over a single root key shared across services.
  • Rotate on a schedule (90 days is a sane default) and immediately on suspected leak.
  • Audit usage in the provider console after rotation to confirm the old key has zero traffic.
  • Set per-key spend limits where the provider supports them, so a leaked key has a bounded blast radius.

Pricing at a glance

Anthropic bills per million input/output tokens, with prompt caching giving a 90% discount on cached input tokens. Sonnet is the price/performance pick for most workloads; Haiku is the cheap-and-fast lane; Opus is the heavy-reasoning tier. The Batch API gives a flat 50% off.

FAQ

What's the difference between an Anthropic API key and a Claude.ai login?
Different products. Claude.ai is the consumer chat. The API key authenticates against the Anthropic platform and is what you'd use from your code.
Why does my key need workspaces?
Workspaces let you set spend caps and rate limits per environment. Production should always be in its own workspace with its own key.
Does Anthropic have a free tier?
Anthropic offers small starter credits on signup. They expire. After that it's pay-as-you-go.
What's the latest Claude model?
As of early 2026, Claude Opus 4.7 is the flagship; Sonnet 4.6 and Haiku 4.5 cover the price/performance and speed tiers. Always pull /v1/models to see the current set tied to your key.
How do I enable prompt caching?
Set cache_control breakpoints in the message content. Cache hits are billed at ~10% of normal input cost.
Why is my key returning 403 even though it's valid?
Workspace permission. The key may be valid but tied to a workspace that has no model access enabled. Check the workspace settings.

Test other providers

Related reading