MiniMax H3 local video generation takes up to 58 minutes per clip
A 10-second 1920×1080 MiniMax H3 T2V render took 58 minutes on an RTX 5090 in ComfyUI (INT8, 20 steps, 24 FPS); a 864×480 render took 15m 50s on an RTX 5080. Users report quality is better at full resolution but want a good H3 upscaler.
2 sources
Daily brief
Get tomorrow's AI brief in your inbox
More stories today
- ChatGPT Work used to install OpenClaw and Ollama, run local model
- Local-first workspace supports Claude Code, Codex, and holaOS
- Open-source RAG framework supports any data source at any scale
- Qwen 3.6 runs locally on Radeon R9700 and multi-Mac agent farms
- Tool compiles knowledge bases and runs parallel research across Claude, Codex, Pi