Paper 2510.26583
Emu3.5: Native Multimodal Models are World Learners
- Published
- Oct 2025
- Research lab
- Independent
- Citations
- 109
- GitHub
- 1.5K stars
01 In brief
Summary
Emu3.5 is a large-scale multimodal world model from BAAI that natively predicts the next state across vision and language.
It is pre-trained end-to-end with a unified next-token prediction objective on over 10 trillion tokens from interleaved vision-language data, primarily derived from internet videos.
The model accepts and generates interleaved vision-language inputs and outputs, and is post-trained with large-scale reinforcement learning to enhance reasoning and generation.
To improve inference efficiency, the authors propose Discrete Diffusion Adaptation (DiDA), which converts token-by-token decoding into bidirectional parallel prediction, accelerating per-image inference by about 20× without sacrificing performance.
Emu3.5 demonstrates strong native multimodal capabilities, including long-horizon vision-language generation, any-to-image (X2I) generation, and complex text-rich image generation.
It also exhibits generalizable world-modeling abilities, enabling spatiotemporally consistent world exploration and open-world embodied manipulation.
In comparisons, Emu3.5 achieves performance comparable to Gemini 2.5 Flash Image (Nano Banana) on image generation and editing tasks and shows superior results on interleaved generation tasks.
The model is open-sourced at https://github.com/baaivision/Emu3.5 to support community research.
02 From the paper
Abstract
We introduce Emu3.5, a large-scale multimodal world model that natively predicts the next state across vision and language. Emu3.5 is pre-trained end-to-end with a unified next-token prediction objective on a corpus of vision-language interleaved data containing over 10 trillion tokens, primarily derived from sequential frames and transcripts of internet videos. The model naturally accepts interleaved vision-language inputs and generates interleaved vision-language outputs. Emu3.5 is further post-trained with large-scale reinforcement learning to enhance multimodal reasoning and generation. To improve inference efficiency, we propose Discrete Diffusion Adaptation (DiDA), which converts token-by-token decoding into bidirectional parallel prediction, accelerating per-image inference by about 20x without sacrificing performance. Emu3.5 exhibits strong native multimodal capabilities, including long-horizon vision-language generation, any-to-image (X2I) generation, and complex text-rich image generation. It also exhibits generalizable world-modeling abilities, enabling spatiotemporally consistent world exploration and open-world embodied manipulation across diverse scenarios and tasks. For comparison, Emu3.5 achieves performance comparable to Gemini 2.5 Flash Image (Nano Banana) on image generation and editing tasks and demonstrates superior results on a suite of interleaved generation tasks. We open-source Emu3.5 at https://github.com/baaivision/Emu3.5 to support community research.