Skip to content

Show-o

One transformer for multimodal understanding and generation

Visit official site

What Show-o is

The model is described as a single transformer for multimodal understanding and generation, developed jointly by Show Lab at the National University of Singapore and ByteDance.

What you can do with it

  • Repair damaged image regions from text guidance
  • Extend an image beyond its original frame
  • Combine text and image into new multimodal content

Getting started

  • Install the required environment and dependencies
  • Download and configure pre-trained model weights
  • Log in to wandb to view inference demo results
  • Run the multimodal understanding inference demo

Pros & cons

✓ What we like

  • Mixed modality generation is included
  • A single transformer covers understanding and output

! What to watch out for

  • Pre-trained weights must be downloaded before inference
  • A wandb account is used for the demo results

FAQ

Who built it?

The material names Show Lab at the National University of Singapore and ByteDance.

What tasks are listed?

Captioning, visual question answering, text-to-image generation, inpainting, outpainting and mixed modality generation.

Last reviewed: 2026-11-15

More AI image generation tools

View all →

How we review