Wan2GP

A local AI media generator with a browser interface for video, images, music and speech, including select models that need just 6 GB of VRAM.

Alternative to:

Wan2GP generates media on your own computer for creators who want to work with several AI models in one browser interface. It builds on Wan-Video/Wan2.1 and focuses on making generation accessible on GPUs with limited memory.

Video and image generation: Create videos with Wan, MiniMax H3, LTX-2 or Hunyuan Video, and images with models including Qwen Image, Krea 2 and Flux.
Music and speech: Generate audio with Ace Step or Stable Audio, produce speech with Qwen3 TTS, add video soundtracks with MMAudio, or replace voices with SeedVC.
Editing and finishing: Prepare inputs with masks, background removal and pose or depth extraction, then improve video outputs with RIFE or FlashVSR; LTX-2.5 also supports transparent video extraction and turning rough viewport animations into finished shots using an appearance reference.
Reusable generation workflows: Queue media jobs, browse outputs in galleries, recover generation settings and save templates; LoRAs and custom checkpoints let you adapt supported models.
Offline task agent: Deepy coordinates generation and handles tasks such as transcription and video splitting offline, with a focus on low VRAM use.

Wan2GP is free to use locally. Select models run with as little as 6 GB of VRAM, though memory needs depend on the model and output. It supports NVIDIA GTX 10XX, RTX 20XX and newer cards, plus AMD RDNA hardware. Quantized checkpoint support includes GGUF, int8, fp8 and Nunchaku. An API lets developers add generation to their own applications.