MiniMax 3-step Turbo LoRA tested: 2-min audio+video gen on RTX 5090
agapes1270 · reddit · 2026-09-14
A Reddit user tested Kijai's new MiniMax H3 Taomate 3-step Turbo LoRA (built on the base MiniMax model), sharing four sample generations with synced audio and video.
- At 0.7MP, generation takes about 2 minutes on an RTX 5090 with 128GB RAM
- Sampling: euler+simple on most clips, ersde+beta on one; shift 12 for video, 3 for audio
- All clips use identical seeds and prompts for comparison; the LoRA (avg rank 19, bf16) is available for download
More from Infra
- Your AI agent would cost 260% more without prompt caching, Arena data shows — arena · 2026-10-30
- Nebius runs auction-driven GPU pricing, lifting rates ~20% amid tight supply to 2028 — BenBajarin · 2026-10-30
- macOS update silently re-adds eMudhra root certs users had removed — steipete · 2026-10-30
- Bittensor's Terza Inference routes one API across miners' models on subnet 103 — const_reborn · 2026-10-30
- SGLang Summit 2026 sells out sponsor booths and Full Passes ahead of Nov 12–13 event — ying11231 · 2026-10-30
- OpenRouter rebuilds provider routing: 46% fewer capacity errors, 16.9% faster TTFT — steipete · 2026-10-30