Open-weights audio-driven avatar video model from Meituan, built on LongCat-Video, with 8-step distilled inference.
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.
LongCat-Video-Avatar 1.5 is an audio-driven human video generation model from Meituan's LongCat team, built on the LongCat-Video foundation model. It supports Audio-Text-to-Video, Audio-Text-Image-to-Video (animating a reference image from audio) and video continuation, with single-stream and multi-stream audio inputs.
Compared with v1, the audio encoder is changed from Wav2Vec2 to Whisper-Large, which the authors say gives smoother lip dynamics. The model is described as stable for long videos with consistent identity, and it generalizes to anime, animal characters, multi-person conversations and object handling. DMD2-based step distillation reduces inference to 8 function evaluations. A YouTube review mentions an int8 version of about 16 GB.
Weights are published on Hugging Face under the MIT license, with a project page and a technical report. It is aimed at developers and creators who run the model locally or on their own infrastructure for use cases such as broadcasting, singing, e-commerce and animation. The model card lists English and Chinese as supported languages.
Watch the moment: the exact second top creators show it. Videos load only when you press play, via youtube-nocookie.com.
Talking avatar generator from a reference image plus audio, more stable and expressive than v1, supports animation styles and multi-person conversation. Weights released; int8 version is 16 GB.
Closest neighbours by meaning, not just by category.
Model · Meituan
Meituan's open-weights 1.6T-parameter multimodal MoE model (48B active) for agentic coding and visual tasks.
Model · ByteDance
ByteDance image-and-audio-driven digital human video model, offered through the BytePlus API.
App · HeyGen
AI video platform for creating avatar-led videos from text, with voice cloning and lip-synced translation in 175+ languages.
Model
Research model for real-time, audio-driven talking-head video generation from a reference image, using one-step inference.
Model
Research model for real-time, talking video agents with a single streaming Transformer; 640×368 output at ~200 ms latency.
Model · Mirage
Avatar video model from Mirage, marketed for identity preservation in AI-generated talking-person videos.
