Key Info

MiniMax-H3 is now optimized for AMD MI355X, achieving up to 26.7x faster inference than SGLang on 8 GPUs, with 5-second video generated in 1.3 seconds.

Highlights

  • Nunchux brings MiniMax-H3 to AMD MI355X with up to 26.7x faster inference than SGLang on 8 GPUs.
  • Streaming generation allows prompt changes while the video plays.
  • One stack optimized across hardware, giving creators a faster feedback loop.