Google releases DiffusionGemma, a diffusion-based text generation model

Google introduces DiffusionGemma, a 26B-parameter open model generating text via diffusion rather than token-by-token, mirroring image generation techniques.

Published 9sem1 sourceNotable

Lire en français

≈ 24sSpeed

The fact

Performance gains: ~1,000 tokens/sec on H100 GPU (4x faster than comparable autoregressive models), but with lower output quality.

Experimental positioning: Google offers it to developers as a research tool, acknowledging the speed-quality tradeoff.

Click the link to read an article on the topic:

Dev.to

Explore this topic

Google Nvidia #diffusion #génération textuelle #LLM #open model #inference

What if you saw the whole news differently?Factae cross-checks hundreds of sources worldwide to keep only the fact, no opinion. Explore the front page.

Auto-synthesis from 1 media source · identified on June 10, 2026

← Back to home

The fact

Read more

Google launches Gemini 3.7 Flash, an optimized version of its AI model

Google ramps up AI projects despite internal challenges

Google Gemini surpasses one billion monthly users