Terms of service

Last updated 11 October 2026.

These terms cover the batchrate batch-inference service at batchrate.ai. By creating an account or sending an API request, you agree to them.

The service

batchrate accepts OpenAI-style batch files and returns model output within a 24 hour completion window. Processing runs on interruptible capacity. A lost worker is retried. We do not promise a specific queue time inside that window, and a batch can end as completed, failed, cancelled, or expired.

Accounts and keys

You are responsible for the API keys you create. Keys are shown once. Revoke a key if it leaks. You must give a real email address and verify it before using the API.

Credits

Usage is prepaid. Credits are added when Stripe confirms payment, not when the browser returns from Checkout. Jobs are rejected when the available balance is below the estimate for that file. Completing a job debits actual input and output tokens at the prices then stored for the model. Unused holds are released. Credits are not cash and are not transferable. Placeholder prices on the site are labeled as such until replaced in config.

Acceptable use

Do not use batchrate to break the law, to attack systems, to generate child sexual content, or to submit data you do not have rights to process. We may cancel jobs and suspend accounts that create operational or legal risk.

Your content

You keep ownership of prompts and outputs. You give batchrate permission to store and process that content to run the job, prevent abuse, and produce the output file. See the privacy page for retention and subprocessors.

Disclaimers

Model output can be wrong. Do not rely on it for medical, legal, or other high-stakes decisions without review. The service is provided as available. To the extent the law allows, batchrate is not liable for indirect or consequential loss, and direct liability is limited to the credits you paid in the three months before the claim.

Contact

Questions about these terms: contact.