header-langage
简体中文
繁體中文
English
Tiếng Việt
한국어
日本語
ภาษาไทย
Türkçe
Scan to Download the APP

HKU Diffusion Model Team Raises Nearly 500 Million Yuan, Backed by Huawei and Shunwei

Beating AI News Flash: Shenzhen-based diffusion language model company DiffuSpace has completed two consecutive funding rounds, raising a total of nearly $500 million. Matrix Partners China, Shunwei Capital, and Legend Capital co-led the rounds, with Huawei Hubble and Horizon Robotics participating as follow-on investors. Multiple media outlets are calling this the largest funding round in the global diffusion language model sector.


DiffuSpace was co-founded in May of this year by University of Hong Kong professor Kong Lingpeng and his doctoral students Gong Sansan and Ye Jiacheng. The team previously jointly launched Dream 7B with Huawei's Noah's Ark Lab and has open-sourced the model weights. It employs diffusion language model (dLLM) technology. Mainstream models like GPT and Claude typically generate text sequentially, whereas diffusion models can process multiple positions simultaneously and then progressively fill in and refine the content.


DiffuSpace is currently training a new model with 30 billion parameters, with plans to release and open-source it soon. The company will focus on advancing code agents, as well as local deployment on devices such as phones, cars, and robots.

举报 Correction/Report
Correction/Report
Submit
Add Library
Visible to myself only
Public
Save
Choose Library
Add Library
Cancel
Finish