Usage and billing

Use TT to check managed agent usage, cloud credits, and run costs. Direct local training, evaluation, and serving do not consume Tuned Tensor credits.

Save your access token once with tt auth login to use these account commands.

Managed agent allowance

tt usage

Managed inference uses an included daily request allowance. The allowance is shared across your account's tokens and CLI instances and resets at 00:00 UTC. Each accepted model request counts, including requests after a tool call and requests that fail or are cancelled. One agent conversation can use several requests.

tt usage shows the limit, remaining requests, request outcomes, and reported token usage. Reported provider cost is informational; managed requests do not debit your cloud credit balance. Token and cost totals include only requests for which that information was received. The coverage fields show how many requests are included.

Cloud credits

tt balance
tt cloud runs report <run-id>

tt balance shows prepaid credits, reservations for work in progress, and recent transactions. Labeling and older hosted training jobs use this ledger. New training runs locally or on your own AWS instance; AWS bills that capacity directly, without TT training credits.

Run tt topup to choose an amount and open Stripe Checkout. The billing dashboard also shows your balance and transactions and lets you manage payments.

Your own provider key

When you configure your own OpenRouter key, OpenRouter handles inference billing and usage reporting. You can choose any model available to your OpenRouter account. Those requests do not use the TT managed allowance.