DiffusionGemma released: text diffusion model achieves 4x faster inference
Original title:DiffusionGemma: 4x faster text generation
AI summary
Google released DiffusionGemma, an experimental open-source model that uses text diffusion technology to generate 256-token blocks in parallel, improving inference speed by up to 4x. The 26B MoE model activates only 3.8B parameters, fits quantized on consumer GPUs with 18GB VRAM, and reaches 1000+ tokens/s on a single H100.
Key Takeaways
- Google released DiffusionGemma,an experimental open-source model that uses text diffusion…
- The 26B MoE model activates only 3.
- 8B parameters,fits quantized on consumer GPUs with 18GB VRAM,and reaches 1000+ tokens/s on a…
Quick Facts
- Published
- Updated
- Source
- Google AI:DEV 作者专属(RSS)
- Category
- Model Releases
- AI score
- 68
Google把扩散模型首次做进大语言模型,生文速度提升4倍,还开源了权重。虽然实验性质量打折,但本地实时交互的开发者终于有个新选项可以玩了。
Related entities
Source & copyright
This article is based on curated titles and AI summaries from the AIHOT public API, with attribution and links back. Original copyright remains with the original publishers and authors; this site does not mirror full texts.
How to Cite
When citing this article, keep the title, date, and permalink, and credit Orange News as the source.
Orange News: DiffusionGemma released: text diffusion model achieves 4x faster inference. July 14, 2026. https://new.nat6.net/en/news/cms3dqexj0aviro3fnuu5mxbg/