Low-VRAM platform whose author says MiniMax H3 runs on as little as 5 GB VRAM at 480p. The creator points to a previous install tutorial.
Open-source local web app for running video, image, audio, music and 3D generative models on low-VRAM GPUs.
Updated EU status checked Last shown
Wan2GP at a glance
- What is Wan2GP?
- Wan2GP is an app. Open-source local web app for running video, image, audio, music and 3D generative models on low-VRAM GPUs.
- Does Wan2GP work in the EU?
- Available in the EU. It is a locally run application distributed via GitHub with no regional restrictions stated. Individual model licenses may differ, but none of the provided material excludes the EU. Last verified 14 Nov 2026.
- How much does Wan2GP cost?
- Open source. Free to use locally, at home or at work, per the README. Individual models carry their own licenses. Licence: Custom / not declared (GitHub reports NOASSERTION).
- Where was Wan2GP shown?
- In 1 creator video by AI Search. Most recently by AI Search on 5 Aug 2026, at 20:17 in “The BEST local AI video generator is here! Minimax H3 tutorial”.
Sources: the vendor's own pages (linked above) and creator videos at the linked timestamps. EU status is information, not legal advice; see how we check.
WanGP, by DeepBeepMeep, is a local web application that bundles many open-source generative models in one interface. It started as a low-VRAM front end for Wan 2.1/2.2 video models and now also covers video (MiniMax H3, LTX-2/2.3/2.5, Hunyuan Video, LongCat, Kandinsky), image (Flux, Qwen Image, Krea 2, Z-Image, HiDream), audio and TTS (Ace Step, Qwen3 TTS, Index TTS, Chatterbox), and 3D models (Hunyuan3D 2.1, TRELLIS.2, TripoSG).
It is built for users with limited GPU memory. The README says select models can run with as little as 4 GB of VRAM, and it supports older Nvidia cards (GTX 10XX, RTX 20XX and newer) and AMD RDNA 2 to 4. It handles quantized checkpoint formats (int8, fp8, gguf, NV FP4, Nunchaku), downloads files suited to the hardware, and supports LoRAs and custom finetunes. Other features include a generation queue, headless batch mode, an API, an MCP interface for external agents, and an offline mode.
It also includes Deepy Prime, a local assistant that plans and runs generations, and experimental interactive "world" models (Hy-Nomas, EVOKE Turbo) that can be explored with the keyboard. Post-processing tools cover upscaling, frame interpolation, soundtrack generation and voice replacement. The README says it is free to use locally. Each underlying model keeps its own license, and the repository license is not clearly declared.
- Runs many video, image, audio, music and 3D models from one browser UI
- Low-VRAM operation, with select models from 4 GB
- Supports Nvidia GTX 10XX/RTX 20XX and newer, plus AMD RDNA 2 to 4
- LoRA and finetune support, queue, headless mode, API and MCP interface
- Deepy Prime local assistant and explorable world models
- Offline mode and upscaling, interpolation and audio post-processing tools
As seen on YouTube
Watch the moment: the exact second top creators show it. Videos load only when you press play, via youtube-nocookie.com.
Similar AI tools
Closest neighbours by meaning, not just by category.
MAGI-2
Model · Sand AI
Open-weights 114B MoE model from Sand AI that generates synchronized video and audio from text or image prompts.
FastH3
Model · Hao AI Lab
Open-weight 4-step sparse-distilled version of MiniMax H3 for fast text-to-video-and-audio generation on Blackwell GPUs.
Seedance 2.0
Model · ByteDance
ByteDance Seed's multimodal video generation model, producing video from text and reference inputs.
WorldPlay 2
Model
Real-time interactive world model with compressed memory and hybrid action and semantic control, built on Wan 2.2.
LTX 2.5 GGUF
Model
Community GGUF-quantized builds of the LTX 2.5 video model for running on low-VRAM GPUs, e.g. in ComfyUI.
Mobile Wan
Model · Qualcomm AI Research
Qualcomm research model that runs a distilled, pruned Wan2.2 5B video diffusion model on a mobile device.
