Google DeepMind releases DiffusionGemma, a model that runs local AI 4x faster

Google DeepMind announced DiffusionGemma, a diffusion model that speeds up local AI execution by four times. Diffusion techniques are primarily used for image generation, but they can also accelerate text output. The model is designed for on‑device use, providing faster performance without relying on remote servers.

Summary written by Kernelia from the original article by Ars Technica — AI. The story and its rights belong to its author.