ByteDance image-and-audio-driven digital human video model, offered through the BytePlus API.
Updated EU status checked Last shown
Sources: the vendor's own pages (linked above) and creator videos at the linked timestamps. EU status is information, not legal advice; see how we check.
OmniHuman 2.0 is ByteDance's digital human video generation model. It produces talking, singing or acting avatar video from a single image plus an audio track and, per the BytePlus page, an optional text prompt. The vendor describes it as a "film-grade" model with highly dynamic motion.
According to a creator review, this version extends single-shot output to up to five minutes (previously about 60 seconds) at 720p. The same review reports a price of $0.07 per second and raises concerns about lip-sync quality and deepfake misuse. These details come from the video and the BytePlus page's meta description, not from technical documentation.
It is accessed as a paid API via BytePlus (ByteDance's enterprise cloud). Use cases listed by the vendor include e-commerce, advertising, brand marketing, short video, film and TV, AI music videos, UGC content and fantasy vlogs. Weights and code are not published, and no paper was found in the provided material.
Watch the moment: the exact second top creators show it. Videos load only when you press play, via youtube-nocookie.com.
Generates a talking, singing or acting video from a single image plus audio, now up to five minutes in one shot (previously 60 seconds) at 720p. Price dropped to 7 cents per second; the creator is skeptical of lip-sync quality and notes deepfake concerns.
Closest neighbours by meaning, not just by category.
Model · Meituan
Open-weights audio-driven avatar video model from Meituan, built on LongCat-Video, with 8-step distilled inference.
App · HeyGen
AI video platform for creating avatar-led videos from text, with voice cloning and lip-synced translation in 175+ languages.
Model · ByteDance
ByteDance Seed's multimodal video generation model, producing video from text and reference inputs.
Model
Research model for real-time, audio-driven talking-head video generation from a reference image, using one-step inference.
Model · ByteDance
ByteDance Seed's video generation model supporting multi-reference inputs and one-take creation.
Model · ByteDance
Unreleased, unnamed ByteDance world model demoed in Beijing for novel camera paths and text-to-world generation.
