Tokens and billing
How usage is counted, where to check balance, and the review bonus.
How tokens are counted
Each request bills input + output:
usage = input_tokens + output_tokens- 1:1 rate - no hidden multipliers per model
- Cached prompt tokens are not billed
- Only tokens actually processed upstream count
Example: 10,000 input + 2,000 output → 12,000 deducted from your key.
Check your balance
Open the dashboard and paste your API key:
https://lawthropic.xyz/dashboard
| Field | Meaning |
|---|---|
| token_limit | Key limit (purchase + bonuses) |
| tokens_used | Spent so far |
| remaining | Left |
| percent_used | Usage percentage |
Grace +2%
When you hit the nominal limit, the key doesn't stop instantly - you get +2% grace.
| Key limit | Grace | Hard cap |
|---|---|---|
| 2,000,000 | +40,000 | 2,040,000 |
| 100,000,000 | +2,000,000 | 102,000,000 |
Grace is a small buffer for final requests, not extra quota.
Out of tokens?
- Check dashboard
- Buy another key or contact the seller
Billing FAQ
Are thinking tokens billed? Yes, if upstream reports them in output_tokens.
Claude Code resends full history every request - normal? Yes. Each call bills full input for that request. Use /compact or start a new chat for long sessions.
Why doesn't remaining match my estimate? Dashboard updates after each completed request. Streaming requests settle when the stream finishes.