# Pricing

Every bill is exactly two lines, and nothing else.

**Your model spend, passed through at cost.** What the model providers charge for your calls, passed straight through with no markup. There is no LLM tax. Change model and you owe us the same.

**$1 per platform-minute, for active work only.** Our charge for the orchestration and compute we actually performed on your behalf.

No seats. No storage. No compute line. No per-feature menu. No base fee. When you use Escape Velocity you are billed for intelligence operations, and for nothing else.

**You pay for active work, never idle.** A platform-minute is execution time with the waiting removed - time spent waiting on a model provider, on an external event, or on you is not billed. If two agents work in parallel, you pay for each agent's active work, but none of either one's waiting.

That is deliberate, and it is about incentives rather than generosity. If waiting counted, we would earn more the longer we sat doing nothing, and a slower provider would pay us better. Charging only for work performed means we earn by doing, and we are motivated to make waits cheap and execution fast.

**Bring your own model keys, and the first line goes away.** Add your own provider keys in Settings and your model calls run on your account, billed by that provider directly rather than by us. The platform-minute charge is unchanged, because the orchestration still happened. Your model spend becomes controllable to zero without leaving the platform.

You pay in prepaid credits, and [how those work](/credit-terms) - buying, spending, automatic top-up, expiry and refunds - is set out separately. The purchase, tax and payment mechanics are in our [Terms of Service](/terms).

Rates can change, and changes apply going forward only. They never change the dollar value of credits you have already bought.