Customer Policies

Terms of Service

These terms govern your use of the Benchwick inference API and the models Benchwick hosts on your behalf.

1. Agreement

These Terms of Service are an agreement between you and Benchwick LLC for use of the Benchwick inference API, together with any order form, usage plan, or written quote that references them. By sending requests to the API you accept these terms.

2. The service

Benchwick serves open-weight models on infrastructure it operates and exposes them through an OpenAI-compatible API. Model availability, context limits, and supported parameters are published at the models endpoint. We may add, update, or retire models; where practical we announce a retirement in advance.

3. Accounts and credentials

You are responsible for keeping API keys confidential and for all usage made with them. Notify us promptly if a key is exposed so it can be rotated. Credentials are issued per account and may not be resold or shared outside your organization without our written consent.

4. Acceptable use

You agree not to use the service to:

  • Violate any applicable law, or infringe the rights of others.
  • Generate content that is unlawful, defamatory, or that infringes third-party rights.
  • Attempt to disrupt, overload, probe, or gain unauthorized access to the service.
  • Circumvent rate limits, quotas, or billing, including through automated key rotation.
  • Resell raw access in a way that misrepresents the underlying provider.

5. Your content

You retain ownership of the prompts, inputs, and other content you submit. You grant Benchwick a limited license to process that content solely to deliver responses and to meter usage. You are responsible for having the rights needed to submit the content, and for reviewing model output before relying on it in production.

6. Output and model limitations

Model output is probabilistic and may be inaccurate, incomplete, or unsuitable for a particular purpose. The service is not a substitute for professional advice. You are responsible for validating output used in your own products.

7. Fees and billing

Usage is metered by input and output tokens at the rates published for each model. Invoicing terms, payment currency, and payment period are set out in the applicable order form or written quote. Amounts are due as invoiced. We may adjust published rates with notice; adjustments do not apply retroactively to usage already invoiced.

8. Rate limits and availability

We apply requests-per-minute and token-per-minute limits per account, and may return a rate-limit response when limits are reached. We target high availability but do not guarantee uninterrupted service unless a written service level is agreed in an order form. Planned maintenance is announced where practical.

9. Suspension and termination

Either party may terminate on written notice. We may suspend access immediately for non-payment, breach of acceptable use, or a security risk to the service or to other customers. On termination you remain liable for usage incurred before the effective date.

10. Disclaimers and liability

Except as expressly stated, the service is provided on an “as is” basis without warranties of any kind, whether express or implied, including fitness for a particular purpose. To the maximum extent permitted by law, Benchwick’s aggregate liability arising from the service is limited to the amounts you paid for the service in the three months preceding the claim. Neither party is liable for indirect, incidental, or consequential damages.

11. Changes to these terms

We may update these terms as the service evolves. The revision date below reflects the current version. Continued use after an update constitutes acceptance of the revised terms.

12. Contact

Questions about these terms, or requests for a signed agreement, go to sales@benchwick.com.

Last updated: November 17, 2026.