AI/ML
Run WAN 2.1 on Azure for High Performance Video Generation
Introduction
Alibaba's Wan 2.1 makes use of Microsoft Azure's GPU virtual machines (VMs) for video production pipelines. This guide describes how it will be deployed on Azure, with an eye towards developers and businesses around the world.
Prerequisites
- Azure Account: With VM access.
- Hardware: NCasT4_v3 (NVIDIA T4, 16GB VRAM).
Deployment Steps
Step 1: Create VM
- Image: Ubuntu 20.04.
- Size: NCasT4_v3.
- Storage: 50GB.
Step 2: Connect
SSH:
ssh -i <key.pem> azureuser@<vm-ip>
Step 3: Install and Download
Setup:
sudo apt update && sudo apt install -y python3-pip gitgit clone https://github.com/Wan-Video/Wan2.1.gitcd Wan2.1huggingface-cli download Wan-AI/Wan2.1-T2V-14B --local-dir ./modelspip3 install -r requirements.txt
Step 4: Run
Generate:
python3 inference.py --model_path ./models/Wan2.1-T2V-14B --prompt "A flying drone" --output drone.mp4
Requirements
- Instance: NCasT4_v3
- VRAM: 16GB
- Cost/Hour: $0.918
Conclusion
As of March 2025, using Wan 2.1 on Azure's NCasT4_v3 VM provides an affordable video pipeline. This configuration, which combines Wan 2.1's open-source capabilities with Azure's scalability, is appropriate for creators and businesses.
Ready to transform your business with our technology solutions? Contact Us today to Leverage Our AI/ML Expertise.
AI/ML