Skip to main content
How we handle your data depends on the endpoint you use.

​
Standard endpoints

Liquid may save prompts and outputs and use them to improve our models and services.

​
ZDR endpoints

For endpoints identified as zero data retention (ZDR), we use request content, including prompts, system instructions, files, and responses, only to process and return the requested response. Other than temporary in-memory processing required to complete the request, we don’t log, store, cache, or otherwise retain request content. Any temporary processing data is deleted when the request is complete. We don’t use request content to train, fine-tune, evaluate, benchmark, or improve our models or services.

​
Usage information

For all endpoints, we may keep basic usage information, such as token counts, latency, errors, and billing data, but not information that can reconstruct a request or response.

​
Terms and privacy

The Privacy Policy explains how we handle personal data, including personal data contained in request content and usage information. Your use is also subject to our Terms of Service and any additional terms that apply to the service, endpoint, or platform you use. If you access a Liquid model through a third-party platform, check the data information and terms shown for that endpoint on the platform.