AI Model Release Tracker China · updated
- 1Releases tracked
- Oct 28Latest · Clover-35B-A3B-Preview
- 100%Open-weight
- HKHong Kong
DiffuSpace is a diffusion language model lab started by members of the University of Hong Kong NLP group (HKU NLP), the team behind the Dream 7B diffusion model released in April 2025. Its first release, Clover-35B-A3B-Preview (October 2026), is an Apache-licensed block diffusion language model derived from Qwen3.6-35B-A3B, with 35B total and about 3B active parameters, a 128K context window and a thinking mode. Weights are on Hugging Face and ModelScope, with online and API access offered on request.
DiffuSpace
1 release latest Oct 28, 2026
-
Clover-35B-A3B-Preview
Apache-licensed block diffusion LLM (35B MoE, ~3B active) built from Qwen3.6-35B-A3B; self-reported 73.6 SWE-bench Verified and 680 tok/s on one GB300.
DiffuSpace FAQ
- What is DiffuSpace's latest AI model?
- DiffuSpace's most recent release in the Startrise tracker is Clover-35B-A3B-Preview, released Oct 28, 2026.
- How many AI models has DiffuSpace released?
- Startrise tracks 1 notable DiffuSpace release since January 2025, from Clover-35B-A3B-Preview (Oct 28, 2026) to Clover-35B-A3B-Preview.
- Does DiffuSpace release open-weight models?
- Yes. 1 of 1 tracked DiffuSpace releases (100%) ship with downloadable weights; the most recent is Clover-35B-A3B-Preview.
- Where is DiffuSpace based?
- DiffuSpace is based in Hong Kong.