HeyGen’s Avatar IV turns digital twins into an uncanny mirror

Recording video content usually means endless retakes, awkward lighting setups, and stiff smiles after an hour in front of the lens.

HeyGen founder Joshua Xu just announced that the platform's Digital Twin feature is now powered by a new underlying model called Avatar IV, and it is available immediately.

Head-to-toe mannerisms

The upgrade focuses heavily on making AI avatars look less like robotic talking heads.

According to Xu, Avatar IV can replicate a user's subtle facial expressions, personal mannerisms, and natural gestures.

Instead of just moving the lips, the new model moves naturally from head to toe. It also analyzes the provided script to deliver lines with the specific tone and cadence you would normally use in real life.

HeyGen is pitching this directly at solo creators, founders, and executives who need to churn out educational videos or social updates without burning hours filming.

Why it matters

For anyone running a personal brand or client education channel, creating lifelike video without picking up a camera is a massive time-saver. If you have hesitated to try digital clones because of that uncanny valley stiffness, this is worth checking out.

Across the industry, avatar tech is shifting from basic lip-syncing toward believable physical presence.

However, we think near-perfect twins also raise the stakes on trust. When an avatar captures your exact posture and quirky mannerisms, audiences can hardly tell what was actually said by you, making potential misuse or brand missteps much harder to untangle.