Plain-AyI
← All projects
AI models & research

VAR (Visual Autoregressive Modeling)

A research method and models that create images by drawing them coarse to fine, one level of detail at a time.

Research-backedNew this week

What it does

VAR is a research project on a new way for AI to create images. Instead of filling in an image pixel by pixel, it starts with a blurry, low-detail version and adds finer and finer detail.

Its makers say this lets GPT-style models, the kind behind chatbots, beat diffusion models, another popular image method, in their tests. They also report that results improve predictably as the models get bigger.

The project shares trained models and code. Its makers have since built text-to-image and text-to-video projects on top of it.

Who it's for

AI researchers and students interested in image generation.

Can I use it without coding?

Developers only. It is research code; running it needs programming and a capable computer. New to this? Start here.

Can I run it?

Cost
FreeNothing to pay or sign up for.
Computer
Works best on a powerful computer or one with a graphics card (GPU)
Privacy
Runs on your own device: your data stays with you
API key
Not needed

Try it

Using Claude Code, Cursor or another AI coding assistant? Paste the prompt and it will walk you through the setup.New to this?

Before you install

  • Only download from the official GitHub page. Copies with similar names exist for many popular projects.
  • License (MIT): Very permissive: free to use, change and share, including for business.
  • More on staying safe

Where people are talking about it