AIAndroid Authority2 days ago
Google's DiffusionGemma trades quality for speed
1 min read
Google released DiffusionGemma, an experimental AI model that generates entire blocks of text at once and refines them, rather than writing word by word. It can be up to 4x faster, hitting 1,000+ tokens per second on NVIDIA H100 hardware. Output quality remains inferior to Gemma 4, making it an experimental tool for developers.
Level
Hype check
Tap to vote and see what everyone thinks.
