MiniMax-H3 on AMD MI355X: 26.7x faster inference, 5s video in 1.3s with prompt steering
junyanz89 · x · 2026-09-24
Nunchux runs MiniMax's MiniMax-H3 video generation model on AMD MI355X, claiming up to 26.7x faster inference than SGLang on 8 GPUs — 5 seconds of video generated in 1.3 seconds. Streaming generation lets users change the prompt while the video plays, steering what happens next. MiniMax amplified the news, calling it model optimization meeting AMD inference engineering. Free access to MiniMax-H3 is coming via a waitlist.
Related event: Nunchux speeds MiniMax-H3 on AMD MI355X by up to 26.7x(2 posts)→
More from Infra
- Leaked memo: Meta infrastructure head Santosh Janardhan to leave in early 2027 — dinabass · 2026-11-12
- Omarchy 4.6 lets you share local AI models across your tailnet, says DHH — alexcovo_eth · 2026-11-12
- Eric Schmidt's foundation funds French NGO fighting against data centers — StephanSturges · 2026-11-12
- Epoch: Open-Weight GLM 5.3 Hacks Six Firefox Versions for ~$1,000 Each; Claude Fable 5.2 Sets ECI Record — Epoch AI · 2026-11-12
- Polymarket prices 28% odds of a state data center moratorium; Torus raises $500M — Polymarket · 2026-11-12
- AWS-wrapper cloud startup shuts down just one month after seed round — thedealdirector · 2026-11-12