WanGP — локальная генерация видео и аудио

★ 9.7K

WanGP is an open-source all-in-one app for running the best open generative models locally: video, images, audio and text-to-speech. Reach for it when you want to generate video from text or images on your own computer without sending data to the cloud, when paid services are too expensive or unavailable, or when your GPU is modest and you need to fit a model into little memory. It includes Wan 2.1/2.2, MiniMax H3, LTX-2/2.3/2.5, Hunyuan Video, Flux 1/2, Krea 2, Qwen Image, Qwen3 TTS, Ace Step and more; quantized formats int8, fp8, gguf and NV FP4; LoRAs and custom checkpoints from Hugging Face and CivitAI; a job queue, a headless command-line mode and the WanGP API for adding generation to your own apps. It also offers a browser UI, a mask editor, background removal, pose and depth extraction, temporal and spatial upsampling (RIFE, FlashVSR) and plug-ins. Some models run with as little as 6 GB of VRAM; Nvidia GTX 10XX and newer and AMD RDNA 2–4 are supported. It suits local content production and experiments with open models; it is not for machines without a compatible GPU. Local use is free.