HeyGen launches Avatar 6 for more natural AI avatar movement

HeyGen has launched Avatar 6, an audio driven AI avatar model that interprets a user's script and footage to generate more natural gestures, expressions and lip sync. The model is available today, November 17, 2026, through the HeyGen platform and API.
- Avatar 6 generates gestures and expressions from speech context and personal footage.
- Users can create avatars with as little as 30 seconds of video.
- The model supports improved lip synchronization and native 2K video output.
- HeyGen requires identity verification and explicit consent for realistic avatars.
- Avatar 6 is available through HeyGen and its API.
Avatar 6 uses speech context and personal footage
Avatar 6 is designed to address repetitive or unnatural movement in AI generated video by linking motion to what a speaker says and how they deliver it. According to HeyGen, the model interprets the context and tone of a user's content, then generates gestures and expressions that reflect the individual's delivery.
HeyGen says users can create an avatar with as little as 30 seconds of video. They can also upload additional footage manually or connect personal YouTube, Instagram or TikTok accounts. The company uses that material to understand how each person naturally looks, moves and expresses themselves.
The model's runtime context window has increased 10x, from approximately 3 to 30 seconds, according to the company. HeyGen says this supports more varied movement based on a user's footage, along with improved lip synchronization and native 2K video output.
Testing and identity controls are central to the launch
In Turing style testing comparing recorded footage with AI generated avatar video, 84% of participants perceived the avatar as equally realistic as the person it represents, according to HeyGen. The company also said Avatar 6 ranked number one in New Measure's Avatar Arena, a blind, head to head test where voters preferred HeyGen's model over competitors.
“Modeling human realism remains one of the hardest challenges in AI video,” said Rong Yan, HeyGen's chief technology officer. He said Avatar 6 makes progress in modeling human looks and movement while giving people greater control over how their identity is represented on screen.
HeyGen requires live identity verification and explicit consent to create realistic avatars of real people. The company said third party platforms can generate photo based and virtual avatars, while realistic videos of real people remain within HeyGen's platform. Existing Personal Avatars will be upgraded to Avatar 6 without requiring users to record new footage.
Avatar 6 expands HeyGen's video, voice and avatar stack
The launch follows HeyGen Video, the company's first general purpose video model, and HeyGen Voice, its in house voice model. HeyGen said HeyGen Voice debuted at number one on Artificial Analysis' leaderboard, while HeyGen Voice and HeyGen Voice Turbo ranked among the top models on Artificial Analysis' Controlled Voice arena.
HeyGen described the releases as a full stack offering spanning video, voice and avatars. The company said its technology is used by over 40 million people across 196 countries, with more than 118 million videos created and adoption across 85% of the Fortune 100. It also said it surpassed $200 million in ARR in June, doubling revenue in eight months.
For users, Avatar 6 is intended to reduce the need to film every video while preserving a more individualized presentation. It is available through HeyGen and its API, with support for select third party platforms. HeyGen's launch positions more natural movement and consent controls as key parts of its effort to make AI generated video suitable for people and businesses that rely on digital content.

