'Coffee is ready' AI UGC clip made with MiniMax H3 on PixVerse
Video model · MiniMax
MiniMax H3
15s max2KAudioOfficial price#1 AA I2V
Base model for HeyGen Video 1.0, Utopai X and Creatify Boreal-H3 (each vendor says so).
Specs
| Version | H3 / H3 Max |
| Released | Jul 2026 (AA listing); open-sourced 2026-08-03; H3 Max Aug 2026 minimax.io ↗ checked 2026-10-17 |
| Max length | 15 s |
| Max resolution | 2K (H3-Regenerate-2K, hosted); 768p open-weights base |
| Audio | Yes |
| X | @MiniMax_AI (also @Hailuo_AI for the app) |
Features
- open weights (33B, community licence)
- text/image/video/audio context
- 32 kHz stereo audio, 11 dialogue languages
- 24 fps
Pricing Official list
| 768p | $0.08/s |
| 2K | $0.13/s |
Where to use it
- MiniMax API; Hugging Face weights minimax.io ↗ checked 2026-10-17
- Higgsfield ($0.13/s list) higgsfield.ai ↗ checked 2026-10-17
Benchmarks
| Board | Rank | Score | Board date | Source |
|---|---|---|---|---|
| AA Text-to-Video v2.1 (768p) | #4 | 1159 | live | artificialanalysis.ai ↗ checked 2026-10-17 |
| AA Text-to-Video v2.1 (H3 Max) | #5 | 1155 | live | artificialanalysis.ai ↗ checked 2026-10-17 |
| AA Image-to-Video | #1 | 1140 | live | artificialanalysis.ai ↗ checked 2026-10-17 |
| LMArena Text-to-Video | #8 | 1460 | 2026-09-21 | lmarena.ai ↗ checked 2026-10-17 |
| LMArena Image-to-Video | #1 | 1495 | 2026-09-21 | lmarena.ai ↗ checked 2026-10-17 |
| DesignArena Image-to-Video (H3 Max) | #2 | 1324 | 2026-10-16 | x.com ↗ checked 2026-10-17 read from chart image |
Use cases
One-prompt UGC shoe product demo for about $1 using MiniMax H3 Max on fal + ElevenLabs
Spec ad for a made-up sticker brand, built with MiniMax H3 on OpenArt
Picsart x MiniMax H3 Challenge, $50k prize pool (selfie / product-image animations)