Utopai X
Released Sep 30, 2026
Utopai Studios' video generation model, used inside its PAI production-intelligence platform for film and television. Filmmakers select it to generate and compare takes for a shot.
- Price per second
- —
- Max clip length
- —
- Max resolution
- —
- Native audio
- Yes
- AA arena Elo
- 1168
- LMArena Elo
- —
- Capabilities
- text to video
- Licence
- Proprietary
Strengths & weaknesses
Strengths
- No. 2 on Artificial Analysis' text-to-video leaderboard with audio (Elo 1168)
- Developer cites reflections, caustics, world understanding and high-level spatial consistency
Weaknesses
- No API available (Artificial Analysis); usable only inside the PAI platform
- No published specs for duration or resolution
releaseDate is the date of the developer's announcement page (Sep 30, 2026); AA lists release as 'Sep 2026'. The developer page gives Elo 1,150 (nine points behind Wan 3.0); AA-Video-T2V v2.1 shows 1168 (2026-10-27), which is used. The AA row carries a MiniMax logo, and third-party coverage (orcarouter.ai, Chinese press) says it is a post-train of MiniMax H3, and Utopai's FAQ confirms it: 'Utopai X is post-trained on MiniMax H3'. nativeAudio is from Utopai's FAQ: 'Utopai X generates video with audio.' PAI is billed usage-based (about page); no per-second price is published. Not on LMArena's video board.