MiniMax H3 beats Wan 2.2 in head-to-head: $0.03 per clip, but invented speech needs fixing
victor_explore · x · 2026-09-25
A solo developer building reels for a quit-vaping app ran a controlled faceoff between Wan 2.2 and MiniMax H3 using identical stills, voice, and captions. H3 (the #1 open video model on both arenas) is sharper and more dynamic, but invents speech — whisper detected words in 80% of clips; pinning the audio track to room tone cut bad clips from 3-11 per reel to 1-2. Cost is identical at $0.03/clip ($0.40/reel) on a rented 4090. H3 is now the default, with Wan 2.2 kept as backup since H3's license excludes the US, EU, UK, and South Korea.
More from Multimodal
- Sophon-Realtime-v1, a full-duplex speech-to-speech model, trends on Hugging Face — adventists-ai · 2026-11-11
- Asked Fable 5.2 to 'go wild' making a game, got a Pong clone — and loved it — ariG23498 · 2026-11-11
- Western town built with Fable 5.2: ray-traced weathering ages every board — ravisparikh · 2026-11-11
- Diffusion Week 2026 releases all 10 lectures on video generation, world models and more — LucaAmb · 2026-11-11
- Running Wan 2.2 on a RTX 3080 12GB: 5B Breaks Down, 14B Works Surprisingly Well — fixxxer2606 · 2026-11-11
- Indie dev teases Z-Image-based model 'on par with latest closed-source models' — darlens13 · 2026-11-11