Type: AI model
DiffusionGemma is a Google DeepMind open-weight model built on the 26B-A4B mixture-of-experts Gemma 4 architecture that generates tokens via discrete diffusion, reaching up to 1000+ tokens per second on a single H100. It is multimodal over text, image, and video inputs and released under Apache 2.0 on Hugging Face. GROUNDING tracks DiffusionGemma’s diffusion-based generation, benchmarks, and deployment.
Recent Updates
- 2026-08-21: mistral.rs: Agentic local LLM inference with Anthropic Messages API support (breakingnewsofficial) · github.com — Agents Tool Use Open Source LLMs Anthropic · OpenAI · Hugging Face EricLBuehler · Qwen3-4B Muse Glimmer-30B · Gemma 4
FAQ
What is DiffusionGemma?
DiffusionGemma is a Google DeepMind open-weight model built on the 26B-A4B mixture-of-experts Gemma 4 architecture that generates tokens via discrete diffusion, reaching up to 1000+ tokens per second on a single H100. It is multimodal over text, image, and video inputs and released under Apache 2.0 on Hugging Face. GROUNDING tracks DiffusionGemma’s diffusion-based generation, benchmarks, and deployment.
What does this page track?
Dated radar mentions, source links, related concepts, and builder-relevant context for DiffusionGemma, collected automatically by GROUNDING.
When was DiffusionGemma last mentioned?
DiffusionGemma was most recently mentioned in a radar update dated 2026-08-21.
Category: Image Generation