AI Matters
Why it matters
Google releases DiffusionGemma for faster text generation
The Decoder Google's DiffusionGemma proves you don't need to train from scratch to build a text diffusion model
Google DeepMind released DiffusionGemma, an experimental model that generates up to 1,500 tokens per second by adapting an existing model rather than training from scratch.
Why it matters
Generating text in parallel rather than word-by-word could significantly speed up AI assistants while reducing the computing power they require.
Latest brief · Aug 9, 2026