Questions, answered
Everything you might want to know about running a self-hosted AI studio on your own hardware. Can't find it here? The team is one message away.
Getting started
What exactly is Aurora?
Aurora is a self-hosted creative studio for AI image and video. Instead of routing your work through someone else's paid API, it orchestrates open backends — ComfyUI, Stable Diffusion, LTX-2 and self-hosted LLMs on Vast.ai GPUs — on infrastructure you control.
Do I need a powerful GPU to use it?
A modern GPU with plenty of VRAM makes the experience smoother, but the studio scales down: lighter checkpoints and shorter clips run on modest cards, while larger video models reward more memory. The GPU Costs page estimates the load of each model before you queue it.
Is there a cloud version I can try without installing anything?
This build is designed to run on your own hardware. You can explore the full interface, the model catalog and the studios here; generation happens once you point Aurora at a running backend.
Models & generation
Which models are supported?
Image generation runs through Stable Diffusion checkpoints and ComfyUI graphs; video runs through LTX-2 and compatible pipelines; text and prompt assistance run through a self-hosted LLM on Vast.ai (vLLM, OpenAI-compatible). New backends can be added through the catalog without rewriting your projects.
Can I bring my own checkpoints and LoRAs?
Yes. Because the pipeline is yours, you can drop in your own checkpoints, LoRAs and ComfyUI workflows and they appear alongside the built-in options.
How do credits work on a self-hosted build?
Credits are a virtual accounting layer so you can budget and compare model cost. On your own hardware there is no per-image fee — you tune the credit values in the catalog to reflect your real electricity and time.
Data & ownership
Where are my generations stored?
On the storage you connect to Aurora. Outputs, prompts and project history stay on your infrastructure unless you deliberately export or publish them.
Do you train on my prompts or images?
No. Aurora does not collect your prompts or outputs to train models. What you make is yours.
Can I use what I create commercially?
Generally yes, subject to the licenses of the specific models and assets you load. Check the license of each checkpoint, LoRA or font you use, and review the Terms for the full picture.
Plans & billing
What do the paid plans include?
Paid tiers unlock convenience features around the studio — priority queueing, larger private libraries, team seats and copilot assistance. The core generation always runs on your own machines.
Can I cancel any time?
Yes. Plans are month to month and you can downgrade or cancel from your account at any point. See the Refund Policy for how partial periods are handled.