How to Find Your Claude AI API Key: A Quick Guide
If you're searching for "how to find claude ai api key," you're probably staring at the Claude.ai chat interface wondering where the developer settings are, or you signed up for something once and can't remember where the key went. The short answer: your Claude API key lives in the Anthropic Console (console.anthropic.com), not in the consumer Claude.ai app. It's a separate product with separate login and separate billing.
The key itself is a string starting with sk-ant-api03- followed by a long random token. You generate it once, copy it immediately (Anthropic won't show it again), and store it somewhere safe like a .env file or a secrets manager. If you've lost it or never saved it, you can't "find" the old one — you have to revoke it and create a new one.
Where the Key Actually Lives
The consumer chat app at claude.ai has no API key section because it's built for conversational use, not programmatic access. The API key is generated in a different product:
- Go to console.anthropic.com and log in (or sign up if you haven't).
- Look for API Keys in the left sidebar or account settings.
- Click Create Key, give it a name (e.g. "production-backend"), and copy the value shown.
- Store it in your environment as
ANTHROPIC_API_KEYor similar — never hardcode it in source files.
If you already created a key months ago and just need to check it exists (not read its value — Anthropic never shows the full key again after creation), the console lists key names, creation dates, and last-used timestamps. You can rotate or revoke from that same screen.
Why You Might Not Find One
A few common reasons people end up here:
- They only used claude.ai, the free/paid chat product, which has no API access built in.
- They confused Claude with a third-party tool that wraps Claude (browser extension, Notion AI, etc.) — those don't expose a raw Anthropic key.
- Someone else on the team created the key and it lives under a workspace they don't have access to. Ask an admin to either share console access or issue you a scoped key.
- The key was rotated or revoked after being accidentally committed to a public repo (GitHub scans for this and Anthropic will auto-revoke leaked keys).
If none of these apply and you genuinely have no console account, the fix is simple: create one at console.anthropic.com, add a payment method, and generate a fresh key. There's no way to "recover" a key from claude.ai because it was never issued there.
Checking a Key Actually Works
Once you have a key, confirm it's valid before wiring it into an app:
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-3-5-sonnet-20241022",
"max_tokens": 100,
"messages": [{"role": "user", "content": "Say hi in five words."}]
}'
A 200 response with a content array means the key is live. A 401 means the key is wrong, revoked, or missing from the request headers.
If You Want to Skip Console Management Entirely
Some teams don't want to manage raw Anthropic keys, workspace billing, or per-developer console access — especially if multiple people or services need Claude access without sharing one root key. That's the gap SubToAPI fills: it sits between your existing Claude access and your codebase, issuing scoped application keys (sub_live_...) instead of handing every developer the master Anthropic key.
Instead of hunting through console settings, you get a key from a single dashboard:
curl https://api.subtoapi.app/v1/messages \
-H "Authorization: Bearer $SUBTOAPI_KEY" \
-H "content-type: application/json" \
-d '{
"model": "claude-3-5-sonnet-20241022",
"max_tokens": 100,
"messages": [{"role": "user", "content": "Say hi in five words."}]
}'
Each teammate or service can get its own sub_live_ key, with usage tracked separately per key instead of pooled under one shared credential. That makes it much easier to answer "where's our API key" six months from now — it's in the SubToAPI dashboard, scoped to whoever needs it, not buried in one person's Anthropic console login.
Setup takes a few minutes: create an account at /signup, generate a key, and follow /docs/quickstart to send your first request. Plans start at Solo (€9/month) for individual use, with Team (€19/seat) and Scale (€49/seat) tiers for shared access — full details on /pricing.
Keeping Your Key Safe Once You Find It
Whichever key you're using — raw Anthropic or a SubToAPI application key — treat it like a password:
- Never commit it to git. Add
.envto.gitignorebefore your first commit, not after. - Use environment variables or a secrets manager in production, not hardcoded strings.
- Rotate keys periodically and immediately if you suspect exposure.
- Scope keys per service or per developer where possible, so a leak doesn't compromise everything at once.
Questions
Does claude.ai (the chat app) have an API key section? No. The consumer chat product and the developer API are separate. API keys are only issued through console.anthropic.com, or through a proxy service like SubToAPI if you're using one.
What does a valid Claude API key look like? Anthropic keys start with sk-ant-api03- followed by a long alphanumeric string. SubToAPI application keys follow the format sub_live_.... Both should be treated as secrets and never exposed client-side.
I can't find my old key — can support look it up for me? No provider will show you a previously created key's full value again, for security reasons. The fix is always to revoke the old key (if it still shows in your dashboard) and generate a new one — see /docs for the SubToAPI equivalent workflow.