Your workstation does the work.
Run supported models on your own hardware. Bring image, video and 3D workflows into one creative workspace.
In the making. Almost in your hands.
Your GPU.
Your ideas.
Your control.
Meet Rawrr: a local-first AI creative platform for images, video and 3D. Make it yours with LoRAs. Keep confidential creative work on your workstation—not in a cloud generation queue.
Apply for early accessClosed beta planned for mid-November 2026.
RAWRR / 001
SMALL DINOSAUR. BIG IDEAS.
SMALL DINOSAUR. BIG IDEAS.A NEW CREATIVE WORKSPACEIMAGESVIDEO3DLoRAs
CREATIVE POWER. CLOSER TO HOME.
Create more.
Keep control.
Local generation. No API bill.
Locally executed models do not need a paid inference API for each generation. Your hardware, electricity, software and model licensing costs still apply.
Your client work. Kept local.
Work with confidential briefs, client assets and unreleased concepts without uploading them to a remote inference service in fully local workflows. Choose what leaves your machine—and when.
Your style, with LoRAs.
Go beyond a generic look. LoRAs are lightweight model adaptations that can bring a distinctive style, subject or visual identity to compatible models. Rawrr’s creative direction puts these personalized workflows alongside local generation.
Rawrr is in development. These benefits describe local workflows—not a promise of zero risk, universal compatibility, or zero cost. This website separately processes beta applications. Explore how local AI works ↗
BEFORE YOU APPLY
A few straight answers.
What is Rawrr?
Rawrr is a local-first AI creative platform in development for image, video and 3D workflows, with a closed beta planned for mid-November 2026. It is designed around creators’ own workstations.
Why focus on LoRAs?
LoRAs let creators adapt compatible models toward a particular style or subject without replacing the entire base model. They can help build a recognizable visual direction across projects. Compatibility, results and usage rights depend on the base model and LoRA; Rawrr’s supported workflows will be confirmed during beta.
Does local AI mean no API costs?
For models running locally, there is no third-party per-generation inference API fee. That does not mean the whole product is free: hardware, power, software and model licences may cost money. Any optional external service has its own pricing. Rawrr pricing has not been announced.
Are my prompts and creative files uploaded?
In a fully local workflow, inference runs on your machine without sending prompts and source assets to a remote inference service. External services or sharing change that boundary. The beta application form sends the information you enter to the project operator via Cloudflare and Gmail; see the privacy notice.
Can Rawrr guarantee no confidential leaks?
No software can guarantee that. Local processing reduces the need to send creative work to a third party, but device security, model and plugin trust, backups and integrations still matter. Do not treat a local-first label as a security certification.
What hardware do I need?
Requirements depend on the model, workflow, resolution and available VRAM. A final supported-hardware matrix has not been published. Tell us your OS, GPU and memory in the application; there is no blanket promise that every device will be supported.
When can I try it?
The closed beta is planned for mid-November 2026. Apply below; selected testers receive an email invitation. Applying is free, does not guarantee access, and is not a software purchase.
Meet the project ↗ · Local AI: privacy, costs and trade-offs ↗
THE FIRST WAVE
Make something.
Help shape Rawrr.
We’re inviting a small group of creators to put Rawrr through its paces. Tell us what you do, what you want to make, and what you’re working with.
Closed beta · Mid-November 2026
Applications are reviewed individually.
Selected testers will receive an email invite.
Learn · Compare · Create
Make local AI work for you.
Practical guides for designers, independent creators and studio directors exploring AI on their own hardware.