Terms of Use
The engine is Apache 2.0 — that licence governs the code. These terms cover the website, the Desktop binary and the services run from here.
The short version
The Rapid-MLX engine is open source under Apache 2.0 — read that licence, it is the one that governs the code. These terms cover the website, the Desktop application binary and the services this site operates. Nothing here takes away a right the Apache licence grants you.
The software
The engine is licensed under Apache 2.0; the full text ships in the repository. The Desktop application is distributed free of charge for use on Apple silicon Macs. You may use it personally or commercially. You may not redistribute a modified build as though it were an official release, or use our name, logo or cheetah to imply that we endorse something we do not.
Models are not ours
Rapid-MLX downloads and runs third-party model weights. Each model carries its own licence and its own acceptable-use terms, and complying with them is your responsibility. Some forbid commercial use; some restrict particular applications. We index and make models convenient to run; we do not grant you any rights in them, and listing a model in our catalogue is not legal advice about it.
Model outputs are generated by those models. We make no claim about their accuracy, and we claim no ownership of what you generate.
Acceptable use
Use of this website and the services it operates is subject to the obvious limits: do not attack or overload the infrastructure, do not attempt to gain access you were not granted, do not submit falsified benchmark results, and do not use the newsletter or submission endpoints to distribute spam or malware. We may rate-limit or block traffic that does.
What you run locally on your own Mac is between you and the model's licence. We have no visibility into it and no ability to police it.
Community Benchmark submissions
Sharing a benchmark run is voluntary. When you do, you agree that the measurement and the hardware description may be published on the public leaderboard under your randomly generated contributor name. We may remove submissions that appear falsified or that would identify an individual.
No warranty
The software and this website are provided “as is”, without warranty of any kind, express or implied, including merchantability, fitness for a particular purpose and non-infringement. This mirrors section 7 of the Apache 2.0 licence.
Running large models will use most of your machine's memory. Doing so is not risk-free: it can cause other applications to be terminated by the operating system. Keep backups of anything you care about.
Limitation of liability
To the maximum extent permitted by law, we are not liable for indirect, incidental, special, consequential or exemplary damages, or for lost profits, data or goodwill, arising out of your use of the software or this website — including damages arising from model outputs. Some jurisdictions do not allow these exclusions, in which case they apply to you only as far as the law permits.
Availability
The website, the model mirror and the leaderboard are provided as a convenience and may change or go away. The engine is open source and runs entirely offline once a model is on disk, which is deliberate: nothing we operate needs to exist for your installation to keep working.
Governing law
These terms are governed by the laws of the State of California, United States, without regard to its conflict-of-law rules.
Changes
We may update these terms. The date below records the last change, and the page's full history is public in the website repository.
Contact
hello@rapidmlx.com
MachineFi Lab, 80 Willow Road, Menlo Park, CA 94025, USA
Last updated 11 September 2026.