Skip to content
AIBRAINWORX / THE INDEPENDENT AI LABRESEARCH · HARDWARE · OPEN INTELLIGENCE
AI News / Model Watch

HeyGen releases Avatar 6 for audio-driven talking video

HeyGen says the hosted model takes a photo and can use a short reference clip to capture more of a speaker’s movement.

Publisher: HeyGen · Model / component: Avatar 6
Availability: Hosted HeyGen platform and API.

Listen to this articleBrowser Voice
Ready
Official HeyGen brand lockup
HeyGen brand kit, used for editorial identification.
Event / context: 2026-11-17. Source check: Nov 19, 2026.

HeyGen made Avatar 6 available through its platform and API on November 17, according to its release announcement. The company’s research report describes audio-driven generation from a photo, with optional reference footage to influence appearance and motion. It shows examples up to 2K.

What the evaluation measures

HeyGen reports that viewers selected the generated clip as the real recording in 42.2 percent of a two-clip comparison with the same audio. The company doubles that selection rate for a separate headline score, so readers comparing systems should look at the underlying test and the question the raters were asked. These results come from HeyGen’s own study. AiBrainWorX did not generate or evaluate clips.

The model is offered through hosted services. For teams making training or product videos, the practical evaluation is to watch an entire clip and check for drifting eye lines, repetitive gestures and changes in appearance between segments. HeyGen says it verifies consent for avatars made from a person’s footage. Creators still need permission to use the likeness and voice in their own material.

HeyGen’s research page includes its examples and describes who appears in them. Readers can use those examples to understand the company’s test design before comparing their own footage.

Primary sources reviewed. No independent hands-on tests.

Follow your curiosity.

Pass the signal.