SVTR 硅谷科技评论SVTR
← Database

DiffuSpace

扩散智能

DiffuSpace, founded in May 2026 in Shenzhen, builds diffusion language models (dLLMs) that use parallel generation, bidirectional modeling and iterative refinement to move beyond token-by-token autoregressive generation.

Request an introWebsite ↗
📍 Shenzhen · South ChinaFounded 2026

Position in SVTRnew

Snapshot 2026-10-21
🔒 Sign in for the exact rank and neighbours

About

DiffuSpace, founded in May 2026 in Shenzhen, builds diffusion language models (dLLMs) that use parallel generation, bidirectional modeling and iterative refinement to move beyond token-by-token autoregressive generation. Its model line covers diffusion language models, code models and vision-language / vision-language-action models. The founding team comes from the HKU NLP Lab, led by Lingpeng Kong, assistant professor of computer science at the University of Hong Kong and co-director of the lab; the team has worked on diffusion language models since 2022 and open-sourced Dream 7B, Dream-Coder 7B, Dream-VL/VLA and DiffuCoder.

★ SVTR RatingSVTR has evaluated this company

Full tier & rationale — an SVTR Intelligence benefit

Capital graph

CompanyInvestorFounders / teamCo-investment (thicker arc = more shared deals, ×N marks the count)Bigger circle = stronger signal (activity / coverage / interest) · ⛶ fullscreen · scroll / pinch to zoom · hover to highlight · drag nodes · click a node to open its profile

Funding history

Comparable companies· by market

This company

  • DiffuSpace

    DiffuSpace, founded in May 2026 in Shenzhen, builds diffusion language models (dLLMs) that use parallel generation, bidirectional modeling and iterative refinement to move beyond token-by-token autoregressive generation. Its model line covers diffusion language models, code models and vision-language / vision-language-action models. The founding team comes from the HKU NLP Lab, led by Lingpeng Kong, assistant professor of computer science at the University of Hong Kong and co-director of the lab; the team has worked on diffusion language models since 2022 and open-sourced Dream 7B, Dream-Coder 7B, Dream-VL/VLA and DiffuCoder.

    Undisclosed · 2026-10

Confirmed rounds only. Matched on subsector, sector, stage and funding scale — each verifiable on the company profiles.

Investors that may fit

Tencent

6 Foundation Models deals in the last 24 months · 6 in South China

Request intro →

5Y Capital

4 Foundation Models deals in the last 24 months · 4 in South China

Request intro →

XUHUI CAPITAL INVESTMENT

3 Foundation Models deals in the last 24 months · 3 in South China

Request intro →

IDG Capital

3 Foundation Models deals in the last 24 months · 3 in South China

Request intro →

Shokz

3 Foundation Models deals in the last 24 months · 3 in South China

Request intro →

Huaqin Technology

3 Foundation Models deals in the last 24 months · 3 in South China

Request intro →

Ranked by each firm’s deals in this sector over the last 24 months, weighted by nearby stage and shared geography. Indicative only — no rating judgment.

Want to reach DiffuSpace — for capital, customers, or partnership? Send a short request; SVTR reviews it and brokers the intro, with both sides verified before you meet.

Request an intro to this companySubscribe to Signal