DeepMind DiffusionGemma: 26B Text-Diffusion Model Self-Corrects Mid-Generation; Genie-3 on Street View
Key Questions
What is DiffusionGemma and who announced it?
DiffusionGemma is a 26B parameter text-diffusion model announced by Raia Hadsell at RAAIS 2026. It can self-correct during generation, challenging autoregressive approaches.
How does DiffusionGemma differ from traditional language models?
It uses a diffusion process rather than autoregressive generation, enabling mid-generation self-correction. This offers potential gains in reasoning and controllability.
What other model was revealed alongside DiffusionGemma?
Genie-3 was also revealed, demonstrated on Street View data. It extends generative capabilities in visual domains.
Raia Hadsell announced DiffusionGemma at RAAIS 2026—a 26B parameter text-diffusion model that can self-correct during generation, challenging autoregressive dominance. Also revealed Genie-3 on Street View. Represents a paradigm shift in language modeling with potential for improved reasoning and controllability.