Ars Technica
·
May 6, 2026 3:44 PM
Google's Gemma 4 AI models get 3x speed boost by predicting future tokens
Up to 3x the speed with no loss of quality—is it too good to be true?
Read at Ars Technica