HeyGen upgrades Avatar IV with prompt-driven gestures

AI avatars usually give themselves away the second their hands stay frozen or their facial expressions feel painted on.

HeyGen announced a major upgrade to Avatar IV. The update introduces dynamic gestures generated directly from your script, along with gesture control driven by text prompts. It also brings hyper-real micro-expressions and exports 1080p video up to 60 seconds long. The feature is already live for all HeyGen accounts.

Directing body language with text

Until now, most avatar videos suffered from the talking-head problem. Presenters either stood completely rigid or looped unnatural hand movements that had nothing to do with what was being said.

Avatar IV tries to bridge that gap by aligning movement with context. It reads your script to trigger sensible gestures automatically, but it also lets you prompt specific actions when you need deliberate emphasis. Paired with micro-expressions, the result looks far less synthetic.

Why it matters

For marketing teams, creators, and founders making social clips or product walkthroughs, this is an immediate workflow booster. You get convincing presenter footage without booking studio time or hiring actors, and the barrier to test different messaging stays practically zero.

We think giving users prompt control over gestures is a clever design choice. While pure generative video models try to predict full scenes end-to-end, avatars need precision. Handing the director's steering wheel to the user prevents awkward algorithmic guesses.

The real trade-off right now is the 60-second cap. It is plenty for social ads and elevator pitches, but anyone looking to build full tutorials or webinars will still have to stitch clips together in post.