Open-source diffusion models for synchronized 5-second video and audio generation, with a Full HD super-resolution add-on.
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.
Kandinsky 6.0 Video is a family of diffusion models from Kandinsky Lab for synchronized text-to-audio-video (T2AV) and image-to-audio-video (TI2AV) generation. It comes in two sizes, Lite (3B parameters per the README) and Pro (29B), and produces 5-second clips with synchronized 44 kHz audio, including lip-sync. A separate plug-in super-resolution model raises output to Full HD (1920×1080).
The native pipeline also supports generation from first and last frames, from an audio file, and from a video file (to drive audio). Distilled checkpoints (PiFlow, 10 steps) are provided for faster inference, and a SPEED spectral progressive diffusion option speeds up the Pro model by about 14%. Integrations exist for Diffusers, ComfyUI, vLLM-omni, SGLang and FastVideo, and the community has released INT8 quantized weights.
The code is released under the MIT license and runs locally on NVIDIA GPUs; the README publishes benchmarks from RTX 5060 Ti 16 GB up to H100. A Hugging Face Space demo of Pro Distill is available. Clips are limited to 5 seconds, and one YouTube reviewer judged output quality decent but behind Minimax. Pro models need substantial GPU time or offloading on consumer cards.
Watch the moment: the exact second top creators show it. Videos load only when you press play, via youtube-nocookie.com.
Open-source video generator with Lite (3.8B) and Pro (29B) variants plus a latent super-resolution model. Limited to 5-second clips and text/image-to-video; judged decent but behind Minimax.
Closest neighbours by meaning, not just by category.
Model · Baidu
Baidu ERNIE team's 6.3B joint audio-video generation model with native audio-visual alignment and timbre-controlled speech.
Model
One- and two-step image generation model distilled from Z-Image-Turbo using text-stream adversarial distillation.
Model
Research model for real-time, audio-driven talking-head video generation from a reference image, using one-step inference.
Model · Nvidia
NVIDIA's 5B and 14B hybrid linear-attention video diffusion transformers generating up to 720p on a single GPU.
Model
3D-aware 360° video diffusion model that generates panoramic videos along user-defined camera paths from sparse panoramas.
Model
Pixel-space text-to-image diffusion model transferred from latent diffusion models, without a VAE, supporting 4K output.
