Local AI. Real trade-offs.

Updated 7 October 2026 · Website and beta applications only

Rawrr is being built as a local-first AI creative platform for images, video and 3D. The goal is simple: make your own workstation the centre of your creative workflow.

What “local” means

A supported model runs on your computer rather than a hosted inference API. For that local workflow, your prompts and source assets do not need to be uploaded to a remote inference service. Downloading models, using external integrations or sharing work are separate network activities.

Costs: no inference API meter, not zero expense

Local execution does not charge a third-party API fee per generation. You still need suitable hardware and electricity, and software or model licences may have costs or restrictions. Rawrr’s pricing is not yet announced. Cloud services, if chosen, have their own terms and charges.

Privacy: a smaller sharing boundary

Keeping inference on-device can reduce third-party exposure of confidential creative work. It does not prevent malware, unsafe plugins, compromised backups or accidental sharing. Evaluate the actual model, workflow and device configuration before using confidential material.

Local versus hosted AI

Who should apply?

Artists, filmmakers, designers, 3D creators and other makers interested in testing local workflows. Bring a real project and tell us about your machine. Hardware support and model-specific requirements will be established through testing, not blanket marketing claims.

Current stage

Closed beta planned for mid-November 2026. No public download or paid plan is available on this site. Apply for early access.

Creative AI guides