MiniMax H3 runs 720p to 1440p on single RTX 4090
Independent developer built auto-scheduled sparse attention for X-MinimaxH3, achieving second-sampling times of 112s/223s/334s for 720p/1080p/1440p on one RTX 4090. Another user tested optimizations on RTX 5090 over 12 hours, reporting efficiency gains without quality loss.
2 sources
Daily brief
Get tomorrow's AI brief in your inbox
More stories today
- Redditor tests AI agents with $1 online task
- AI agents need their own identity before a gateway
- Claude Max users find default $200K spend limit
- TTFT-First Benchmark Ranks Lowest-Latency Voice and Realtime Agent APIs
- AI training demand causes Mac Mini shortages