HeyGen Video — a scene-generating text-to-video model, not an avatar — launched from as little as $0.01 per second and leads Artificial Analysis' audio-off text-to-video board in blind human testing.
2026-10-25 · by Moe Ameen
HeyGen built its name — and a reported $200 million in annual recurring revenue — on avatar video: type a script, pick a presenter, get a talking head. HeyGen Video, the model it introduced in late September 2026, is a deliberate turn away from that. It is the company's first general-purpose video model, and it makes no avatar at all: from a single text prompt it generates a whole scene — the picture, the dialogue, and the sound effects — in one pass. HeyGen says it went up against today's leading video models in a blind test on Artificial Analysis and ranked first in text-to-video.
The ranking is real but narrower than the headline. Artificial Analysis scores video models by Elo from blind pairwise votes, where people pick the better of two clips generated from the same prompt without knowing which model made each. HeyGen Video leads Artificial Analysis' silent text-to-video board — the one that judges the picture with audio turned off — with an Elo around 1144. On the main board that includes generated audio it sits in the top tier but not first, trailing Alibaba's Wan 3.0 and Utopai X. So "#1" is specifically the audio-off, text-to-video result, and the margins between the leaders are small.
On specs, the API (model ID heygen-video-1) does text-to-video, image-to-video, and reference-to-video, outputs 480p up to 2K as MP4 with generated audio, and runs clips from about 5 seconds up to 15–30 seconds depending on the mode, in styles from live action to anime to claymation. HeyGen priced it on a 50%-off launch promotion through October: about $0.01 per second at 480p and about $0.015 per second at 768p, with audio; standard rates return to roughly $0.02 and $0.03 per second respectively in November, and 1080p/2K bills at about three times the 768p rate. The HeyGen app is free to start with a couple of clips a day. HeyGen's own developer docs say the model is built on MiniMax's open-weight H3 base and post-trained by HeyGen, rather than a from-scratch build. Treat the exact prices, ranking, and date as a snapshot and confirm them against HeyGen's own pages before you budget around them.
The strategic read is that an identity-first avatar company now has a competitive general-purpose scene generator in the same bracket as Google's Veo, Kuaishou's Kling, Alibaba's Wan, and ByteDance's Seedance — and it is pricing it aggressively.
Here is what a creator can do with this today. Generate a handful of scene clips with HeyGen Video — an establishing shot, a product beat, a mood cut — for pennies each, then bring them into [Kompozy](/) as source footage. Kompozy's [Clipped Shorts](/glossary/clipped-short) takes an uploaded video and cuts it into vertical, word-synced captioned shorts, which is the step HeyGen Video skips: its generated audio does nothing for the majority of feeds that play muted. From there one idea fans out into a [Carousel](/glossary/hyperframes), quote graphics, per-platform text posts, a blog, and a newsletter, all held to your [Persona Brief](/glossary/persona-brief) so the copy stays on voice.
The split is clean: HeyGen Video renders the raw scene, Kompozy turns a pile of raw scenes into a published, on-brand feed. And the avatar lane HeyGen is known for is already inside Kompozy — its [Persona Shorts](/glossary/persona-shorts) and [Persona HeyGen](/ai-tools/heygen) formats generate talking-head video from your own persona pool — so you can intercut a HeyGen Video b-roll scene with an avatar segment and let [Autopilot](/glossary/autopilot) schedule the whole set across the eight social platforms plus blog and email behind a per-post review. The model makes clips; Kompozy makes the calendar.
HeyGen Video (model ID heygen-video-1) is HeyGen's first general-purpose video model. Unlike its avatar products, it generates an entire scene — picture, dialogue, and sound effects — from a single text prompt, and also supports image-to-video and reference-to-video. It launched in late September 2026 via HeyGen's API and app.
It leads Artificial Analysis' silent text-to-video leaderboard — the board that judges the video with audio turned off — based on blind pairwise human-preference votes, with an Elo around 1144. On the main board that includes generated audio it ranks in the top tier but not first, behind Wan 3.0 and Utopai X. The "#1" claim is specifically the audio-off text-to-video result.
HeyGen set a 50%-off launch price through October: about $0.01 per second at 480p and about $0.015 per second at 768p, both with audio. Standard rates return to roughly $0.02 and $0.03 per second respectively in November, with 1080p/2K billed at about three times the 768p rate. The HeyGen app is free to start with a couple of clips a day. Confirm current rates on HeyGen's pages before budgeting.
HeyGen Video outputs a short clip; it does not caption it for muted feeds, reframe it per platform, or publish it. Kompozy does: upload the clip to its Clipped Shorts format for captioned vertical cuts, repurpose the idea into carousels, text posts, a blog, and a newsletter in your brand voice, and schedule it across eight social platforms plus blog and email.