Transfusion: Predict the Next Token and Diffuse Images with One Multi-Modal Model

Chunting Zhou, Lili Yu, Arun Babu, Kushal Tirumala, Michihiro Yasunaga +5 more
2/12/2026

No official code found. Best implementation: lucidrains/transfusion-pytorch (1,356 stars, MIT, last commit 2026-01-27). 3 more below.

Semantic Scholar

Code Implementations(4)

lucidrains/transfusion-pytorchCommunity72%
MIT

Pytorch implementation of Transfusion, "Predict the Next Token and Diffuse Images with One Multi-Modal Model", from MetaAI

1,35671PythonAug 23, 20248 months agoMIT
artificial-intelligenceattentiondeep-learningflow-matchingmulti-modal+1 more
lavinal712/TransfusionCommunity70%
MIT

Transfusion: Predict the Next Token and Diffuse Images with One Multi-Modal Model

10Shell, PythonJun 14, 20258 months agoMIT
VachanVY/Transfusion.torchCommunity64%
PyTorchMIT

PyTorch Implementation of Transfusion: Predict the Next Token and Diffuse Images with One Multi-Modal Model

285PythonAug 24, 20242 years agoMIT
aiartificial-intelligencedeep-learningmultimodal-large-language-modelsmultimodal-transformer+4 more
Ryukijano/TransfusionCommunity61%
MIT

About Pytorch implementation of Transfusion, "Predict the Next Token and Diffuse Images with One Multi-Modal Model", from MetaAI

10Python, Jupyter NotebookFeb 10, 20251 years agoMIT

Cite this paper

@article{zhou2026transfusion,
  title  = {Transfusion: Predict the Next Token and Diffuse Images with One Multi-Modal Model},
  author = {Chunting Zhou and Lili Yu and Arun Babu and Kushal Tirumala and Michihiro Yasunaga and Leonid Shamis and Jacob Kahn and Xuezhe Ma and Luke S. Zettlemoyer and Omer Levy},
  year   = {2026},
  url    = {https://api.semanticscholar.org/CorpusID:fe29071f8db5c333fcced241f70fb6f82e8452e6},
  journal = {ICLR 2025 2025}
}

Discussion