# MiniMax-H3 获 VC-Attention 低比特加速

- 来源：MiniMax (official) (@MiniMax_AI)
- 发布时间：2026-09-18 11:43
- AIHOT 分数：41
- AIHOT 链接：https://aihot.news/items/cmu6fgbcd0mmkrofj3706doi4
- 原文链接：https://x.com/MiniMax_AI/status/2100792780494889194

## AI 摘要

Nunchux AI 与 MIT、CMU、UC Berkeley、斯坦福及 NVIDIA 研究者合作推出 VC-Attention，为 MiniMax-H3 带来免训练低比特注意力加速：在 B200 上比 FlashAttention-4 快 1.6×、B300 上快 1.5×，保真度优于 SageAttention2。

## 正文

Three paths to faster video attention: compute the same interactions more efficiently, compute fewer in full, or change how information is mixed. Here’s a visual guide. 👇

Thanks to Nunchux AI and collaborators for VC-Attention, bringing training-free low-bit acceleration to MiniMax-H3, with better fidelity than SageAttention2 in the B200 evaluation.

The approach balances speed and fidelity: V-Smooth reduces value quantization error, while ExpCast-FP8 makes softmax faster through approximation.

Excited to see the community keep building on H3. Could combining low-bit computation with sparse methods like Sol-Attn push efficiency further? We’re looking forward to seeing that explored.

### 引用推文

> Nunchux AI：Introducing VC-Attention: fast and accurate low-bit attention without retraining. On MiniMax-H3, VC-Attention speeds up attention by 1.6× on B200 and 1.5× on B3...