#3030th of 34 in Open modelsFor image, video and design, video and animation
wan2.1
Open-weight video generation models with PyTorch inference code for text and image input
- Stars
- 17.3k
- License
- Apache-2.0
- Last commit
- Mar 2026
- Language
- Python
15,483 on 31 Mar 202617,263 on 20 Oct 2026
From GitHub's daily star counts.
Checked on GitHub.
Overview
Wan2.1 is a set of open-weight video diffusion models (T2V 1.3B and 14B, I2V 14B, first-last-frame 14B, VACE 1.3B and 14B) with a generate.py CLI, Gradio demos and multi-GPU inference via FSDP and xDiT. The T2V-1.3B model needs 8.19 GB VRAM and makes a 5-second 480P clip on an RTX 4090 in about 4 minutes. Weights are on Hugging Face and ModelScope, and Diffusers and ComfyUI integrations exist.
Who it is for: Engineers running local video generation on NVIDIA GPUs
Strengths
- T2V-1.3B runs in 8.19 GB VRAM on consumer GPUs, 480P
- Covers text-to-video, image-to-video, first-last-frame, and VACE editing in one repo
- Multi-GPU inference with FSDP and xDiT Ulysses/Ring parallelism
- Diffusers and ComfyUI integrations; Apache-2.0 weights on Hugging Face and ModelScope
Weaknesses
- 14B models need far more VRAM; the README gives no figure
- 1.3B model is less stable at 720P; 480P recommended
- Prompt extension needs a Dashscope API key or a local Qwen model
- Diffusers multi-GPU and FLF2V Diffusers/ComfyUI support are unchecked in the todo list
What it needs
- GPU required
- Needs PyTorch >= 2.4.0, Hugging Face Hub or ModelScope for weights, xfuser >= 0.4.1 for multi-GPU, Dashscope API or local Qwen (optional, prompt extension)
- Models: Wan2.1-T2V-1.3B, Wan2.1-T2V-14B, Wan2.1-I2V-14B-480P/720P, Wan2.1-FLF2V-14B-720P, Wan2.1-VACE-1.3B/14B
Also in Open models
See all 34| Rank | Project | Score |
|---|---|---|
| 1 | llmfitRust tool that checks which LLMs fit your RAM, CPU and GPU | 87 out of 100 |
| 2 | layaSingle-pass multilingual classifier for typed choice, score and yes/no decisions | 84 out of 100 |
| 3 | timesfmPretrained time-series foundation model for zero-shot forecasting from Google Research | 75 out of 100 |
| #4 | speechPyTorch toolkit for training and deploying ASR, TTS and speech LLM models | 70 out of 100 |
| #5 | yueLyrics-to-song model that writes an editable melody and chord score first | 65 out of 100 |