RankingsDesignKorean version of this postView as Markdown
AI video models: the blind vote top 10
The ten AI models that make video with audio best, by blind human votes on the Artificial Analysis Video Arena. Updated every month.
- 1Wan 3.0Alibaba · Aug 2026 · $12.00 per minute1172Official site
- 2Utopai XUtopai Studios · Sep 2026 · No API1169Official site
- 3MiniMax H3MiniMax · Jul 2026 · $4.80 per minute1160Official site
- 4FLUX 3 MaxBlack Forest Labs · Oct 2026 · $4.35 per minute1156Official site
- 5Dreamina Seedance 2.5ByteDance Seed · Jul 2026 · $34.12 per minute1155Official site
- 6HeyGen Video 1HeyGen · Oct 2026 · $2.40 per minute1153Official site
- 7MiniMax H3 MaxMiniMax · Aug 2026 · $9.60 per minute1148Official site
- 8FLUX 3Black Forest Labs · Jul 2026 · $17.40 per minute1139Official site
- 9Dreamina Seedance 2.0ByteDance Seed · Mar 2026 · $22.45 per minute1136Official site
- 10Gemini Omni Flash 1.1Google · Aug 2026 · $9.12 per minute1129Official site
In blind votes between two videos made from the same prompt, first place goes to Wan 3.0 with an Elo of 1172. Its interval overlaps with Utopai X (1169) in second and MiniMax H3 (1160) in third, so the gaps at the top are not clear. Six models from first to sixth sit between 1153 and 1172, and Gemini Omni Flash 1.1 in tenth is at 1129.
Price per minute varies a lot. On the makers' APIs, HeyGen Video 1 in sixth is the lowest at $2.40 and Dreamina Seedance 2.5 in fifth the highest at $34.12. Utopai X in second has no API, so it cannot be called directly.
This ranks models, not apps or services. The same model gives different results and prices depending on the service and settings, and other evaluators may put a different model first. The source gives no date, so the date is the day this ranking was made.
If you have tried one of these video models, tell us in the comments.