HeyGen: heygen-video-1

HeyGen Video is an AI video generation model released by HeyGen in September 2026. It generates scene visuals, dialogue and sound effects in a single pass, turning a prompt written like a film shot list — covering subject and action, camera, look, sound and constraints — into a finished clip. The model accepts text, image and reference video input and delivers video, dialogue audio and sound effect output. It offers four visual styles (claymation, live action, anime and screen print), supports 16:9, 9:16 and 1:1 formats, and renders at resolutions from 480p to 2K. Seed values and webhooks are supported for controlled, automated workflows. HeyGen Video draws on HeyGen's avatar technology, with 700+ stock avatars and custom-built digital twins, and supports 175+ languages and dialects. In September 2026, professional voice cloning was added to the API, training a voice from 20+ minutes of a single speaker's recording. HyperFrames lets AI agents create video using HTML with deterministic rendering, frame capture, video playback and preview parity. As of 2026, the model ranked #1 on the Artificial Analysis video leaderboard, and it can be integrated flexibly via the OpenRouter API by changing the model slug. The model is suited to personalized sales outreach, consistent social media production, brand video ads through AI product placement, multilingual localization of course materials within 24 hours, efficient training material production, and interactive tutors, coaches and support agents. Customer results illustrate the speed gains: Happy Cats localized 56 course videos into 7 languages without reshooting, School of AI grew subscribers from 10,000 to 2,000,000 in 8 months, and Tim Stacey cut video production from 2 hours to 10 minutes.