WanGP
Local AI video generator for modest GPUs (6 GB VRAM is enough). Supports Wan 2.1/2.2, Hunyuan Video, LTX Video, Flux, Qwen Image, Z-Image and MMAudio. Gradio web UI with masks, pose/depth extraction and prompt enhancement built in.
Strengths
- Runs on older GPUs (RTX 10xx/20xx) and AMD RDNA 2+, 6 GB VRAM minimum
- Wide catalog: Wan, Hunyuan, LTX, Flux, Qwen, Z-Image, MMAudio, auto downloads
- Full Gradio UI: mask editing, pose/depth, prompt enhancer, video browser
- One-click installers (.bat/.sh), Pinokio and Docker to simplify setup
Limitations
- Heavy setup: Python deps + multi-GB model downloads on first launch
- Many parameters exposed, beginners easily overwhelmed by presets
- Resolution and parameter constraints vary per model
Best for
- Creators iterating on video generation without burning cloud credits
- Local, private usage, prompts and assets never leave the machine
- Comparative benchmarking of open-source models (Wan, Hunyuan, LTX, Flux) in one UI