All tasks / Comparisons
MiniMax H3 vs Utopai X
Compared only on public benchmark results, across the 1 task where both are listed: they share a tier in 1.
MiniMax H3
MiniMax's omni-modal video model with native stereo audio, up to 2K and 15 s; weights released, successor to Hailuo 2.3.
- Vendor
- MiniMax
- Weights
- Open: run it yourself
- Released
- 30 Jul 2026
- API price
- $0.08 per second of video (768p with audio, $4.80/min as listed on the AA text-to-video board ($5.40/min for image-to-video))
- License
- MiniMax H3 Community License Agreement (README links an application form for USA/EU/UK/South Korea)
- Last verified
- 18 Oct 2026
Utopai X
Utopai Studios' film-and-TV video model with audio, offered only inside its PAI production platform (no public API).
- Vendor
- Utopai Studios
- Weights
- Closed: API and apps only
- Released
- 30 Sep 2026
- License
- proprietary
- Last verified
- 19 Oct 2026
Task by task
| Task | MiniMax H3 | Utopai X | Higher tier |
|---|---|---|---|
| Video generation Models | S1,159 on Artificial Analysis Video Arena: Text to Video (AA-Video-T2V v2.1, with audio) · Local | S1,169 on Artificial Analysis Video Arena: Text to Video (AA-Video-T2V v2.1, with audio) · Cloud | Same tier |
A tier spans a range of scores, and small gaps on a leaderboard are often within its margin of error, so only a different tier counts as ahead. How we rank.
Same benchmark, both results
| Benchmark | MiniMax H3 | Utopai X | Snapshot |
|---|---|---|---|
| Artificial Analysis Video Arena: Text to Video (AA-Video-T2V v2.1, with audio) Elo (pairwise human preference), higher is better | 1,159 | 1,169 | 19 Oct 2026 |