# baaivision/Emu3.5

Native Multimodal Models are World Learners

Repository: https://github.com/baaivision/Emu3.5
Canonical: https://ross.abutalabs.com/products/emu35
Language: Python
License: Apache-2.0
License Family: permissive
Last push: 2025-12-30T15:36:41+00:00

## Health v2 (maintenance only)
Score: 43/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 59, release rhythm 35, longevity 22
- inputs: {"age_days": 308, "days_push": 246, "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 1547, forks 69 (observed 2026-08-28T04:05:01.747465+00:00)

## What it is
Emu3.5 is BAAI's native multimodal foundation model that jointly predicts next states across vision and language, trained on 10T+ interleaved video-text tokens. The repo provides model weights, inference code (including vLLM offline inference), and tooling for image generation, editing, and interleaved vision-language generation.

## Use cases
- generate images from text or any modality input
- edit images with natural language instructions
- generate interleaved image-and-text sequences
- run multimodal model inference with vLLM
- explore world modeling and embodied manipulation scenarios
- benchmark against commercial image generation models

## When to choose
- you need an open-weights multimodal model for image generation and editing
- you want interleaved vision-language generation without modality adapters
- you need fast inference via discrete diffusion parallel decoding
- you are researching unified next-token world modeling

## When to avoid
- you only need text-only LLM inference
- you lack GPU resources for large model inference
- you need a production-ready hosted service rather than self-run models

## Facets
- artifact type: library
- maturity: active
- function: machine-learning, deep-learning, llm-inference, image-processing, llm-training
- domain: artificial-intelligence, large-language-models, image-processing, machine-learning
- platform: python
- tags: multimodal, world-model, image-generation, vision-language-model, diffusion, any-to-image, gpu, linux

## Member repositories
- baaivision/Emu3.5 (main) score 43

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:01.747465+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 2026-08-30T04:30:28.429713+00:00, confidence not recorded.
  - readme: https://github.com/baaivision/Emu3.5 (fetched 2026-08-28T04:05:01.747465+00:00, sha 518d7ba5c466)
- Data as of 2026-08-30T08:39:29.467469+00:00.