HeyGen's first general-purpose video model tops Artificial Analysis' silent text-to-video leaderboard
What happened? AI video startup HeyGen announced on 22 October 2026 that Video 1, its first general-purpose video model, ranked first for text to video in a blind test run by Artificial AnalysisDealroom has a profile for this one. Try Dealroom →, where evaluators compare clips from two models generated from the same prompt without knowing which is which.
The fine print: Artificial Analysis now runs two text-to-video boards. On the silent board, HeyGen Video 1 leads with an Elo score of 1,144, ahead of Alibaba's Wan 3.0 (1,134) and MiniMax H3 (1,120). On the board that scores video with generated audio, it ranks third (1,160) behind Wan 3.0 and Utopai X, with overlapping confidence ranges. Artificial Analysis lists its API price at $2.40 per minute, among the cheapest of the top models.
What it does: HeyGen says Video 1 is built on MiniMax H3 and post-trained by HeyGen. It turns a prompt, an image or reference clips into 5 to 15 second scenes with dialogue, room tone and effects generated in the same pass, at 480p to 2K, from $0.01 a second. Inside HeyGen it can be combined with the company's AI avatars, or digital twins, which have been its core product.
Why it matters: HeyGen, which passed $100 million in annual recurring revenue in October 2025, built its business on AI avatar video. Moving into general-purpose scene generation puts it up against Alibaba, ByteDance, Google, Black Forest Labs and MiniMax itself. Investor Sarah Guo, founder of Conviction, reacted: "you can actually just win on model".
Sources: HeyGen on X · HeyGen Video 1 · Artificial Analysis text-to-video leaderboard · Sarah Guo on X
Image credit: HeyGen / Artificial Analysis
