# VectorSpaceLab/OmniGen2

OmniGen2: Exploration to Advanced Multimodal Generation. https://arxiv.org/abs/2506.18871

Repository: https://github.com/VectorSpaceLab/OmniGen2
Canonical: https://ross.abutalabs.com/products/omnigen2
Language: Jupyter Notebook
License: Apache-2.0
License Family: permissive
Last push: 2026-03-20T07:21:13+00:00

## Health v2 (maintenance only)
Score: 52/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 73, release rhythm 35, longevity 32
- inputs: {"age_days": 453, "days_push": 166, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 4112, forks 32 (observed 2026-08-28T04:08:35.748601+00:00)

## What it is
OmniGen2 is an open-source unified multimodal generation model supporting text-to-image generation, instruction-guided image editing, and in-context generation, released with training code, datasets, and a benchmark. It also ships EditScore, a family of reward models for evaluating and improving instruction-guided image editing via reinforcement learning.

## Use cases
- generate images from text prompts
- edit images with natural language instructions
- fine-tune an image generation model on custom data
- run a unified multimodal image generation model locally
- evaluate image editing quality with a reward model
- use an image editing model in ComfyUI workflows

## When to choose
- you need an open-source, Apache-2.0 licensed model for both text-to-image generation and instruction-based editing
- you want to fine-tune or do RL on an image editing model with released training code and datasets
- you want ComfyUI integration or inference speedups like TeaCache and TaylorSeer

## When to avoid
- you need production-grade image generation without GPU hardware
- you only need simple image manipulation like cropping or filtering rather than generative editing
- you need video or audio generation

## Facets
- artifact type: library
- maturity: active
- function: image-processing, machine-learning, deep-learning, llm-inference
- domain: artificial-intelligence, image-processing, computer-vision, deep-learning
- platform: python, cross-platform
- tags: text-to-image, image-editing, diffusion-model, multimodal-generation, instruction-guided-editing, reward-model, comfyui, fine-tuning, gpu

## Member repositories
- VectorSpaceLab/OmniGen2 (main) score 52

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:08:35.748601+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T18:23:11.746890+00:00, confidence not recorded.
  - readme: https://github.com/VectorSpaceLab/OmniGen2 (fetched 2026-08-28T04:08:35.748601+00:00, sha b250d9ce0f52)
- Data as of 2026-08-30T08:39:29.467469+00:00.