{"id":612570,"date":"2026-06-10T16:47:08","date_gmt":"2026-06-10T16:47:08","guid":{"rendered":"https:\/\/Blockchain.News\/news\/nvidia-google-deepmind-diffusiongemma-local-ai"},"modified":"2026-06-10T16:47:08","modified_gmt":"2026-06-10T16:47:08","slug":"nvidia-powers-google-deepminds-diffusiongemma-for-high-speed-ai","status":"publish","type":"post","link":"https:\/\/e-bitco.in\/index.php\/2026\/06\/10\/nvidia-powers-google-deepminds-diffusiongemma-for-high-speed-ai\/","title":{"rendered":"NVIDIA Powers Google DeepMind’s DiffusionGemma for High-Speed AI"},"content":{"rendered":"
\n

Luisa Crawford<\/a> Jun 10, 2026 16:47<\/span> <\/p>\n

NVIDIA optimizes Google DeepMind’s DiffusionGemma for blazing-fast local AI text generation, leveraging RTX GPUs and DGX systems.<\/p>\n

\"NVIDIA <\/a> <\/figure>\n

Google DeepMind\u2019s latest AI<\/a> model, DiffusionGemma, promises to redefine local AI text generation with NVIDIA\u2019s GPU optimizations. Announced on June 10, 2026, DiffusionGemma is built on Google\u2019s Gemma 4 architecture and optimized to run on NVIDIA\u2019s RTX GPUs, RTX PRO platform, and DGX Spark systems. By leveraging NVIDIA\u2019s hardware, DiffusionGemma delivers up to 4x faster text generation compared to traditional large language models (LLMs).<\/p>\n

Unlike conventional autoregressive models that generate text one token at a time, DiffusionGemma uses a parallel processing approach, denoising up to 256 tokens per step. This makes it uniquely suited for latency-sensitive applications such as chatbots, agentic workflows, and on-device AI assistants. NVIDIA\u2019s Tensor Cores and CUDA stack enable this parallelism, maximizing GPU efficiency and cutting response times significantly.<\/p>\n

A New Approach to Text Generation<\/h2>\n

The DiffusionGemma model represents a departure from traditional transformer-based LLMs. It integrates diffusion modeling\u2014commonly used in image and video generation\u2014into text synthesis. By refining entire blocks of text in parallel, the model achieves speeds of up to 1,000 tokens per second on a single NVIDIA H100 Tensor Core GPU. On DGX Spark systems, it delivers up to 150 tokens per second, outperforming autoregressive models in single-user scenarios.<\/p>\n

DiffusionGemma\u2019s architecture builds on Gemma 4, a 26-billion-parameter mixture-of-experts model that activates just 3.8 billion parameters per step, balancing performance with efficiency. The model’s open-weight design, released under an Apache 2.0 license, supports local deployment without requiring cloud-based resources or per-token costs.<\/p>\n

NVIDIA’s Performance Boost<\/h2>\n

Optimized for NVIDIA\u2019s ecosystem, DiffusionGemma is tailored to run efficiently across various platforms:<\/p>\n