Official repo for VGen: a holistic video generation ecosystem for video generation building on diffusion models

Krea Realtime 14B. An open-source realtime AI video model.

[ICLR'2026] Scale-wise Distillation of Diffusion Models

[CVPR2024 Highlight] VBench - We Evaluate Video Generation

Helios: Real Real-Time Long Video Generation Model

[ICLR 2026] LongLive: Real-time Interactive Long Video Generation

Wan: Open and Advanced Large-Scale Video Generative Models

Matrix-Game 3.0: Real-Time and Streaming Interactive World Model with Long-Horizon Memory

Advancing Open-source World Models

NVIDIA FastGen: Fast Generation from Diffusion Models

TurboDiffusion: 100–200× Acceleration for Video Diffusion Models

Implementation of <Streaming Autoregressive Video Generation via Diagonal Distillation> in ICLR 2026

FACM: Flow-Anchored Consistency Models

SANA: Efficient High-Resolution Image Synthesis with Linear Diffusion Transformer

Official codebase for "Self Forcing: Bridging Training and Inference in Autoregressive Video Diffusion" (NeurIPS 2025 Spotlight)

A real-time streaming conversational video system that transforms text interactions into continuous, high-fidelity video responses using autoregressive diffusion.

[ICLR 2026] rCM: SOTA JVP-Based Diffusion Distillation & Few-Step Video Generation & Scaling Up sCM/MeanFlow

Implementation of "Live Avatar: Streaming Real-time Audio-Driven Avatar Generation with Infinite Length"

Official implementation of "PersonaBooth: Personalized Text-to-Motion Generation (CVPR 2025)"

PerMo (PersonaMotion) Dataset

The official implementation of work "AToM: Aligning Text-to-Motion Model at Event-Level with GPT-4Vision Reward".

Nonisotropic Gaussian Diffusion for Realistic 3D Human Motion Prediction: SkeletonDiffusion, a latent diffusion model with attention graph architecture

The homepage of LongCat-Video-Avatar

TTS model capable of streaming conversational audio in realtime.

A high-throughput and memory-efficient inference and serving engine for LLMs

Qwen3 is the large language model series developed by Qwen team, Alibaba Cloud.

HY-Motion model for 3D human motion or 3D character animation generation.

[NeurIPS 2025] Wan-Move: Motion-controllable Video Generation via Latent Trajectory Guidance