Back to AIBriefs
LaunchAI Models

DiffusionGemma: Google's open model generates text in parallel, up to 4x faster

DiffusionGemma: Google's open model generates text in parallel, up to 4x fasterNVIDIA Developer Blog

Generates up to 1000 tokens/sec on single NVIDIA H100, 700+ tokens/sec on RTX 5090. The 26B MoE model (3.8B active) is released under Apache 2.0 and fits within 18GB VRAM when quantized.

DiffusionGemma: 4x faster text generation8 days ago
NVIDIA Accelerates Google DeepMind’s DiffusionGemma for Local AI8 days agoMichael Fukuyama
Awesome to see this innovation in text diffusion. DiffusionGemma is lightning fast, 4x faster than...6 days agoDemis Hassabis
RT @mervenoyann: DiffusionGemma is out 🔥 it's compute-bound so 4x faster compared to other Gemma-4...8 days agoHugging Face
DiffusionGemma is our new experimental open model with up to 4x faster output on dedicated GPUs....8 days agoGoogle DeepMind
DiffusionGemma8 days agoSimon Willison
Google released DiffusionGemma, a new open model with up to 4x faster output! > Instead of...8 days agoTestingCatalog News
Google's DiffusionGemma generates 256 tokens in parallel and self-corrects as it goes7 days ago
Google’s DiffusionGemma is 4x faster than its other Gemma models8 days agoFrederic Lardinois
Google's latest DiffusionGemma open AI model comes with a 4x speed boost8 days agoRyan Whitwam
Google AI Releases DiffusionGemma, a 26B MoE Open Model Using Text Diffusion for Up to 4x Faster Generation8 days agoAsif Razzaq
DiffusionGemma: The Developer Guide- Google Developers Blog8 days agotevlonDiscuss
nvidia/diffusiongemma-26B-A4B-it-NVFP4 · Hugging Face8 days agopmttyjiDiscuss
Diffusion Gemma is 4x faster, but makes 6x more mistakes!6 days agogladkosDiscuss
·
Jun 10, 4:16 PM
DiffusionGemma: Google's open model generates text in parallel, up to 4x faster — AIBriefs