Terms of Use

Last updated: June 27, 2026

By using this website ("the Site"), you agree to these terms.

What the Site does

The Site estimates the memory (RAM/VRAM) and approximate performance needed to run AI models locally, based on live data from a public model repository and on computed estimates.

Accuracy & "as is"

• Model sizes and parameter counts are read from public sources and are believed to be accurate, but are provided without warranty.
• Memory requirements include estimates (KV cache, system margin) and the speed figures (tokens/s) are rough estimates bounded by memory bandwidth — they are indicative only and will differ from real-world results depending on your backend, drivers, and settings.
• "Fits in VRAM / Offload / Insufficient" verdicts are guidance, not guarantees.

No liability

The Site is provided "as is" and "as available". To the maximum extent permitted by law, we are not liable for any loss or damage arising from the use of, or reliance on, the information provided.

Intellectual property

Model names, trademarks, and logos belong to their respective owners. Links to and data about third-party models are provided for informational purposes. Downloads and run commands point to the original model repositories, governed by each model's own license.

Acceptable use

Do not use the Site in any way that disrupts its operation or attempts to overload its services.

Governing law

These terms are governed by the laws of France.

This document is provided for transparency and is not legal advice; please review and adapt it to your situation.