Nunchux AI

@nunchuxai

Building the frontier of multimodal inference for image, video, and world models.

Nunchux says it has been working on AMD Instinct GPUs to make multimodal inference faster and more efficient.

2 Sources, 19d ago, first seen 19d ago

Nunchux says it generated five seconds of video in 1.3 seconds and claims up to 26.7× faster inference than SGLang on eight GPUs.

2 Sources, 34d ago, first seen 34d ago

NunchuxAI says VC-Attention speeds up attention on MiniMax-H3 by 1.6× on B200 and 1.5× on B300 over FlashAttention-4, without retraining.

7 Sources, 41d ago, first seen 41d ago

Fellow academics from MIT reply with praise for the NunchuxAI release.

16 Sources, 54d ago, first seen 54d ago

Featured in