Wan2GP

Wan2GP brings video, image, music and speech generation to your own computer, with particular attention to GPUs with limited memory. It's for creators who want several media models in one browser interface. The project builds on Wan-Video/Wan2.1.

Select models run with as little as 6 GB of VRAM. Hardware support includes older NVIDIA GTX 10XX and RTX 20XX cards as well as newer NVIDIA GPUs and AMD RDNA hardware. It runs on Windows, Linux and macOS, with Docker also available. Quantized model support includes GGUF, int8, fp8 and Nunchaku, while LoRAs and custom checkpoints let you adapt generation to a particular style or subject.

Video choices include Wan, LTX-2, MiniMax H3 and Hunyuan Video. Qwen Image and Flux cover image generation and editing; Qwen3 TTS handles speech, while Ace Step and Stable Audio cover music and audio. Editing tools include masks, inpainting, outpainting and pose or depth controls. RIFE and FlashVSR improve finished video, and MMAudio and SeedVC add soundtracks or replace voices.

A generation queue handles batches, and project workspaces keep imported and generated media together. Galleries let you reuse outputs and recover their generation settings. Deepy is an offline AI agent that can coordinate generation, transcribe recordings and split videos. Optional remote providers, including Codex and Claude Code, connect Deepy and the prompt enhancer to external services. Other devices can access the same running server through a browser, and an API lets developers add generation to their own apps.