HeyGen’s first general-purpose video model, Video 1, ranks first on Artificial Analysis’ silent video board and fourth on its main board

HeyGen, known for AI avatar videos, said on X on October 22 that HeyGen Video 1, its first general-purpose video model, ranked first in text-to-video in a blind test on Artificial Analysis. The model generates short scenes from text, an image or reference images. According to HeyGen, it produces the picture, dialogue and sound effects of a scene in one pass.

Per its API documentation on OpenRouter, clips run 5 to 15 seconds at 480p, 768p or 2K, in six aspect ratios from 21:9 to 9:16. HeyGen’s launch price on its API is $0.01 per second through October. The model itself isn’t new: it has been listed on OpenRouter since September 30. What was new in October was the ranking.

As of October 27, Artificial Analysis’ leaderboard, based on blind votes, shows a different picture. On the main text-to-video board (AA-Video-T2V v2.1), HeyGen Video 1 is fourth with an Elo score of 1,157, behind Alibaba’s Wan 3.0 (1,174), Utopai X (1,168) and MiniMax H3 (1,158). On the separate silent video board, HeyGen Video 1 leads with 1,143, ahead of Wan 3.0 (1,133). The gaps at the top are small, and the rankings can shift as new votes come in.