扩散智能 releases Clover-35B-A3B-Preview diffusion language model for coding and agent tasks
刚刚，扩散模型杀进Agent主场！350亿参数，速度比Jev快一倍
- Who
- Diffusion Intelligence
- What
- released the Clover-35B-A3B-Preview diffusion language model
- When
- Quote
就在刚刚， 扩散智能 发布了全新的 扩散语言模型 —— Clover-35B-A3B-Preview 。
What happened
扩散智能 has released Clover-35B-A3B-Preview, a diffusion language model with 35 billion total parameters, approximately 3 billion active parameters per token, and support for thinking mode. It uses block diffusion, combining parallel updates within blocks with compressed historical context. In team-reported WindTunnel results covering 49 tasks with 3 attempts each, Clover recorded a median execution time of 1.52 seconds and a 98.6% success rate, compared with 3.21 seconds and 95.9% for Jev + Mercury 2.5. Access: https://diffuspace.ai/register
Why it matters
The reported WindTunnel comparison pairs execution time with task success rates, providing a more useful basis for assessing Clover’s speed advantage than isolated generation demos.
Read next
See the full storyDiffusion Intelligence Releases Clover-35B-A3B-PreviewKeep up with AI, one email a day
Model releases, price changes and compute developments, with original sources. Free, confirmed by email, and easy to unsubscribe. Read a past issue · 2026-11-02