Wan AI Video Generator
Alibaba's Wan series is one of the most versatile open-source video generation families available. From the lightweight Wan 2.1 to the flagship Wan 2.6 with 15-second HD generation, run every version available on ComfyUI Web — no GPU, no setup.
Wan AI on ComfyUI Web
Run Wan models directly in your browser. Pick a version to start generating.
Wan 2.6 Image to Video Flash
Transform images into dynamic videos up to 15 seconds with Alibaba Wan 2.6 Flash. Supports 720p/1080p resolution, single or multi-shot modes, and optional audio generation.
Wan 2.5 Image to Video Fast
Alibaba Wan 2.5-fast generates high-quality videos from a single image with faster inference. Supports 720p/1080p and 5s/10s durations, optional audio guidance, and prompt expansion.
Wan 2.2 Image to Video Fast
A very fast and cheap optimized version of Wan 2.2 A14B image-to-video model. Transform static images into dynamic video sequences with smooth transitions and high-quality motion generation.
Wan 2.2 Animate
A unified model for character animation and replacement with holistic movement and expression replication. Transform static images into dynamic video sequences with smooth character animation and precise expression transfer.
Wan 2.1 Mocha
End-to-end video character replacement system that swaps a character in a source video with a new character from reference images. No need for explicit structural guidance like pose or depth maps.
Why Use Wan AI
Image to Video
Upload any photo and transform it into a dynamic video with natural motion, realistic physics, and camera movement. Wan 2.5 and 2.6 support 720p and 1080p output.
Text to Video
Describe your scene in natural language and generate original video clips. Control camera angles, lighting, motion style, and subject behavior through your prompt.
Character Animation
Wan 2.2 Animate transforms static images into dynamic character animations with smooth movement and expression transfer — no pose or depth maps needed.
5 Model Versions
Access Wan 2.1, 2.2, 2.5, and 2.6 from one platform. Each version offers different trade-offs between speed, quality, and capabilities.
100% Cloud-Based
No GPU required, no Python setup, no local ComfyUI installation. Our cloud infrastructure runs the models for you on any device with a browser.
trial-credit tier Available
Start generating videos with no credit card. Free accounts receive credits that refresh regularly. Paid plans start at $9/month for heavier usage.
How to Use Wan AI on ComfyUI Web
Generate your first Wan AI video in under two minutes.
Choose a Wan Version
Pick the Wan model that matches your needs. Wan 2.6 for maximum quality (15s, 1080p), Wan 2.5 for fast image-to-video, or Wan 2.2 Animate for character animation.
Upload or Write a Prompt
For image-to-video, upload a source photo. For text-to-video, describe the scene, camera movement, and style. Example: 'camera slowly orbits around subject, cinematic lighting'.
Generate & Download
Click generate and wait 30-120 seconds. Preview the result, adjust your prompt if needed, and download in MP4 format.
What Is Wan AI?
Wan is a family of open-source video generation models developed by Alibaba's Tongyi Lab. First released as Wan 2.1 in early 2025, the series has rapidly evolved through four major versions — each pushing the boundaries of what open-source video AI can achieve.
Unlike closed-source competitors like Sora or Runway Gen-3, Wan models are freely available under permissive licenses. This means anyone can run them locally through ComfyUI, or use cloud platforms like ComfyUI Web to generate videos without any hardware investment.
Wan Model Versions Compared
Wan 2.1 introduced the foundational architecture with the Mocha variant specializing in character replacement — swapping characters in source videos using reference images without explicit structural guidance like pose maps.
Wan 2.2 expanded the lineup with a dedicated Animate model for character animation and an optimized image-to-video pipeline. The 14B parameter A14B variant delivers strong motion quality at reasonable generation speeds.
Wan 2.5 focused on production speed, offering a 'fast' inference mode that significantly reduces generation time while maintaining visual quality. It supports 720p and 1080p output at 5 or 10 seconds.
Wan 2.6, released December 2025, is the most capable version. It generates up to 15-second HD videos at 1080p, supports multi-lens narrative with automatic storyboarding, and introduces the R2V (Reference-to-Video) model for preserving character appearance and voice across scenes.
Technical Specifications
Wan models are available in 1.3B and 14B parameter sizes. The 14B variant produces higher-quality output with better motion coherence, while the 1.3B version offers faster generation suitable for rapid prototyping.
Supported aspect ratios include 16:9, 9:16, 1:1, 4:3, and 3:4. Resolution options span 720p to 1080p depending on the model version. Wan 2.6 adds native audio-visual synchronization with support for audio-driven video generation.
All Wan models accept text prompts (text-to-video) and image inputs (image-to-video). Wan 2.2 Animate additionally accepts static character images for animation, and Wan 2.1 Mocha accepts character reference images for replacement.
Wan AI vs Closed-Source Alternatives
Wan competes directly with Sora (OpenAI), Runway Gen-3, Kling (Kuaishou), and Pika. While these closed-source models offer polished interfaces, Wan provides comparable quality with full transparency and no vendor lock-in.
On ComfyUI Web, you get the best of both worlds: the openness and flexibility of Wan models with the convenience of a cloud-hosted platform. No GPU purchase needed, no complex setup, and trial credits to get started.
Frequently Asked Questions
- Is Wan AI free to use?
- Yes. Wan is open-source and free to run locally. On ComfyUI Web, you get trial credits to generate videos. Paid plans start at $9/month for more credits.
- What is the difference between Wan 2.5 and Wan 2.6?
- Wan 2.6 supports up to 15-second videos at 1080p (vs 10s for Wan 2.5), adds multi-lens narrative with automatic storyboarding, and introduces Reference-to-Video for character consistency across scenes.
- What resolution does Wan AI support?
- Wan supports 720p and 1080p output. Aspect ratios include 16:9, 9:16, 1:1, 4:3, and 3:4. Video duration ranges from 5 to 15 seconds depending on the model version.
- Do I need a GPU to use Wan AI?
- Not on ComfyUI Web. Our cloud GPUs handle all inference. You only need a web browser. For local use, Wan 14B requires approximately 24GB VRAM.
- Can Wan AI generate videos from text only?
- Yes. Wan supports both text-to-video and image-to-video generation. Describe your scene in natural language and the model generates original video content.
- What is Wan 2.2 Animate used for?
- Wan 2.2 Animate transforms static character images into dynamic video sequences with smooth animation and expression transfer. It requires no pose or depth maps — just a source image.
- How long does Wan video generation take?
- On ComfyUI Web, typical generation takes 30-120 seconds depending on the model version, resolution, and video length. Wan 2.5 Fast is optimized for quicker results.
- Is Wan AI open source?
- Yes. All Wan models are released under permissive open-source licenses by Alibaba's Tongyi Lab. You can download weights from Hugging Face or run them through ComfyUI.