Billing & quotas
How usage is measured, where to check your balance and how limits work.
TokenMoo is usage based: you pay for the tokens you send and receive, not a subscription.
How usage is measured
| Item | Description |
|---|---|
| Input tokens | Everything in messages (including system) plus prompt overhead |
| Output tokens | Generated content, including reasoning where the model emits it |
| Model rate | Each model has its own price per unit; see Model Square |
Input and output are counted separately, and the charged amount is shown in the quota unit used by the console.
Where to look
| Page | What it shows |
|---|---|
| Model Square | Model list with prices and groups |
| Wallet | Current balance, top-up options and spend history |
| Usage Logs | Per-request detail: model, tokens, quota charged, latency |
Top-ups
Recharge from Wallet. After payment the balance is credited automatically; if a top-up does not appear, check the order status on the same page before contacting support.
Quotas and rate limits
- Your account belongs to a group that determines which models are available and at what rate.
- Requests over the concurrency or rate limit return
429; retry with exponential backoff. - When the available quota reaches zero, requests return
402until the balance is topped up.
Keeping costs predictable
- Set
max_tokenson interactive requests to bound output size. - Use smaller models for routine work and larger models only where needed.
- Watch Usage Logs after adding a new feature to catch unexpected token growth early.