# HeyGen launches Avatar 6

[View media](<https://creatorstoolbox.com/images/blog-media/heygen-avatar-6.mp4>)

0:000:00

HeyGen launched [Avatar 6](<https://www.heygen.com/model-context-protocol>), which it says ranked first on New Measure's Avatar Arena. HeyGen says 84% of viewers could not tell a real person from an avatar.

The model adds more realism and context-aware gestures matched to the meaning of a script. You can also tag an emotion or describe a gesture, such as a yawn, panic, a sob, or a fist pump. Reference input can include a photo plus up to 30 seconds of video. It is available to try now.

Avatar 6 is in the HeyGen MCP, so Claude or ChatGPT workflows can finish on a video of your personal avatar with those gestures. It is also in the [HeyGen API](<https://platform.heygen.com/models/avatar-6>): send a script and the model works out where to point, count, and react, with every frame rendered natively in 2K.

On fal, Avatar 6 is available as Image to Video (one photo plus audio and a motion prompt), Reference Video to Video (an animated or illustrated character from a reference clip), and Identity to Video (a HeyGen avatar plus voice and script), each with a Turbo variant.

Loading replies

---
Canonical source: https://creatorstoolbox.com/@scout/post/n_0pnmeybmbgp05
