Summary
According to a post on X by @ModelScope2022, DiffuSpace has released Clover-35B-A3B-Preview, a block-diffusion model built on Qwen3.6-35B-A3B. The model has about 3B active parameters per token and a 128K context window, and it is licensed under Apache-2.0. The post says it claims the top spot among open diffusion language models on its reported benchmarks, and that it ships with Transformers, SGLang and llama.cpp support. The available snippet is truncated, so other details are not covered here.
AI-assisted summary based on the listed source.
What happened
Clover-35B-A3B-Preview is a block-diffusion model built on Qwen3.6-35B-A3B, with about 3B active parameters per token and a 128K context window. It is Apache-2.0 licensed, claims the top spot among open diffusion language models on its reported benchmarks, and ships with Transformers, SGLang and llama.cpp support....
Why it matters
Developers looking for diffusion-based alternatives to conventional language models now have another open option to test, with Apache-2.0 licensing and support across several common inference tools. Teams considering self-hosted deployment may want to check whether the claimed benchmark results hold up on their own workloads. Independent evaluations and early adoption of the released model are the next things to watch.
Signal Intelligence
Signal Strength 95%
Technical label WATCH
Public Interest 63
Category RESEARCH
Reader Depth TECHNICAL
Signal Strength reflects source quality, relevance, freshness and evidence. Public Interest helps organize discovery; it is not proof of truth.
Public Interest components
Recognizable Entity Score 60
Practical Impact Score 82
Novelty Interest Score 94
Consequence Score 34
Curiosity Score 0
Shareability Score 73
Why this is here
VQV surfaced this signal because it is recent, relevant to Open Source LLMs, connected to Scobleizer News Pulse (X).
