Claude AI API Key: How to Get and Use One
A Claude AI API key is the credential that lets your application authenticate with Anthropic's API and send requests to Claude models programmatically. It's a secret string, typically starting with sk-ant-, that you generate in the Anthropic Console and pass in the x-api-key header of every request you make.
If you're looking for this key because you want to build something — a chatbot, an internal tool, a script that automates writing or code review — the short version is: sign up for an Anthropic Console account, create an API key under organization settings, add billing credits, and start making requests. The rest of this article walks through each step, plus the practical issues people run into: rate limits, key security, team access, and what to do if you already pay for Claude Pro but don't want to manage a separate API bill.
How to Get a Claude AI API Key
- Go to the Anthropic Console (console.anthropic.com) and create an account, separate from a consumer Claude.ai login.
- Verify your email and, if required, provide organization details.
- Navigate to API Keys in the settings panel.
- Click Create Key, name it something identifiable (e.g.
prod-backend,staging-worker), and copy it immediately — most consoles only show the full key once. - Add a payment method and load credits, since the API is billed separately from any Claude.ai subscription.
Once you have the key, a basic request looks like this:
curl https://api.anthropic.com/v1/messages \
-H "x-api-key: $ANTHROPIC_API_KEY" \
-H "anthropic-version: 2023-06-01" \
-H "content-type: application/json" \
-d '{
"model": "claude-sonnet-4-5",
"max_tokens": 1024,
"messages": [{"role": "user", "content": "Explain what an API key is."}]
}'The key authenticates the request; without a valid x-api-key header, every call returns a 401.
Where Your API Key Actually Lives
Console-generated keys are tied to an Anthropic organization, not to a personal Claude subscription. That distinction trips people up constantly: paying for Claude Pro gives you higher usage limits inside the chat interface, but it does not give you an API key or API access. The API is a separate product with its own billing, its own rate limits, and its own console.
This matters for planning. If your team already has Claude access through work (Pro, Team, or Enterprise seats) and now needs programmatic access too, you're not upgrading anything — you're setting up an entirely new account with new credits and new keys.
Keeping Your Key Secure
Treat an Anthropic API key like a password with billing attached, because that's effectively what it is:
- Never commit it to source control. Use environment variables (
ANTHROPIC_API_KEY) and a.envfile that's git-ignored. - Never expose it client-side. If you put a raw API key in frontend JavaScript, anyone who opens dev tools can extract it and run up your bill. All API calls should go through a backend you control.
- Rotate keys periodically, especially after a contractor or team member leaves, or if a key was ever logged accidentally.
- Use separate keys per environment (dev, staging, prod) so you can revoke one without breaking the others.
- Set spend limits where the console allows it, so a bug that loops API calls doesn't turn into a surprise invoice.
Common Problems With Direct API Keys
A few patterns show up repeatedly once teams start using the raw Anthropic API directly:
No usage visibility per feature. If five different parts of your product call Claude with the same key, you can't easily tell which feature is driving cost or which one just broke.
No seat-based access control. The Anthropic Console doesn't give you a lightweight way to hand a scoped, revocable key to each developer or client project — you're managing raw keys manually.
Manual rate-limit handling. You're responsible for backoff, retries, and streaming response handling yourself, on every project.
This is the gap SubToAPI fills. Instead of juggling raw Anthropic keys across environments and teammates, you generate application-scoped keys (sub_live_...) from one dashboard, each with its own usage metadata, and route them through a single HTTPS endpoint that supports streaming and tool use. It sits on top of your existing Claude access rather than replacing it — you're still talking to Claude, just through a layer built for teams and apps rather than a single console account.
A request through SubToAPI looks almost identical to a native Anthropic call:
curl https://api.subtoapi.app/v1/messages \
-H "Authorization: Bearer $SUBTOAPI_KEY" \
-H "content-type: application/json" \
-d '{
"model": "claude-sonnet-4-5",
"max_tokens": 1024,
"messages": [{"role": "user", "content": "Summarize this in two sentences."}]
}'The difference is what's around it: per-key usage tracking, team seats, and a dashboard instead of raw console settings. Plans start with a free trial at /signup, and pricing runs from Solo (€9) to Team (€19/seat) to Scale (€49/seat) — see /pricing for details. The /docs/quickstart page walks through getting your first key, and /docs/messages and /docs/streaming cover the request formats for standard and streamed responses.
Choosing Between a Raw Key and a Managed Layer
If you're a solo developer prototyping something small, a direct Anthropic API key is fine — sign up, generate the key, start building. If you're shipping a product with multiple developers, multiple environments, or paying clients who each need their own access, managing raw keys manually becomes a maintenance burden fast. That's the point where a layer with per-app keys, usage metadata, and seat management earns its cost.
Either way, the underlying model access is the same Claude API — the decision is really about how much key and team management you want to build yourself versus get out of the box.
FAQs
Does a Claude Pro subscription include an API key? No. Claude Pro (and Team/Enterprise chat plans) are for the Claude.ai interface only. API access requires a separate Anthropic Console account with its own billing and its own key.
Where do I find my Anthropic API key after creating it? The full key is shown once, right after creation, in the Console's API Keys section. If you lose it, you can't retrieve the same key again — you have to revoke it and generate a new one.
Can I use one API key across multiple apps or team members? Technically yes, but it's poor practice — you lose the ability to track usage per app or revoke access for one person without breaking everyone else. Separate keys per app or environment, or a platform like /docs that issues scoped keys, solves this cleanly.