Urgent.News

600+ sources. One page. See who else covered it.

Editions

More in Tech

multi-token prediction (MTP) support

  • Gemma 4, using Ollama 0.31, shows faster multi-token prediction (MTP)
  • MTP uses smaller draft model to generate and verify tokens quickly
  • Dynamic draft length adjustment maximizes tokens per second generation

More from Tuesday 11 August →