# Tileward pricing: start free, pay when the workload grows

Model usage is pay as you go, from $0.06 per million input tokens. Every account starts with a
$25 credit and no card, and includes Tileward Models, Tileward Context, Tileward Documents and Tileward Governance.

Page: https://tileward.com/pricing/

## Pay as you go

- Each model is billed at its own input and output rate per million tokens; the voice model is billed by the minute. Every model's rates are in [the model library](https://tileward.com/model-library/) ([markdown](https://tileward.com/model-library/index.md)).
- Models that are ready to serve are in the estimator too, at the rate the library gives them, with a way to book time with an engineer: [talk to an engineer about a model that is ready to serve](https://calendar.app.google/nt8a67yTAWrbB44o6).
- Usage draws on account credit. At zero credit the API returns 402 rather than billing you.
- Estimate a month by hand: input tokens times the input rate, plus output tokens times the output rate. The estimator on the page can also apply the 72.1% billable-token cut measured for Tileward Context on six multi-agent problems ([the measurement](https://tileward.com/research/where-context-loses/)); your own threads will differ, and prompt caching narrows it.

## What each product saves

- [Models](https://tileward.com/models/#calculator): fewer cards to hold the same model.
- [Context](https://tileward.com/context/#calculator): fewer tokens resent on every call.
- [Governance](https://tileward.com/governance/#calculator): requests that never reach the model.

## Enterprise, on-prem, or your own VPC

Committed-use pricing, the request-rate limit, deployment in your own VPC or fully air-gapped, SSO and SCIM, and SIEM export are
arranged by contract. Bring one month of usage and an engineer will size it:
[book 30 minutes](https://calendar.app.google/hg2DzFTxKJgwFWm19), or email hello@tileward.com.

## Related

- [Start free](https://app.tileward.com/?join=explore)