Wan 2.5 (preview) API nodes are now live in ComfyUI! A big leap forward for image & video generation 🎵 Audio-Visual Sync (voices, ASMR, music, effects) 🎬 10s Videos (richer storytelling) ✨ 1080P cinematic quality 🔊 Audio conditioning as input Examples 👇
Wan 2.5 Unlimited Video & Image Generator - Multimodal AI With Built-In Audio Sync
Alibaba's cutting-edge multimodal generative AI model supports text-to-video, image-to-video, and text-to-image generation. It delivers crisp high-quality 1080p output, synchronized audio synthesis, flexible 5-10 second duration options, and full multilingual prompt support for a wide range of creative use cases.
Public
WAN 2.5 YouTube Videos
Watch community demonstrations and reviews showcasing WAN 2.5's powerful video generation capabilities
Wan 2.5 Popular Reviews on X
See what people are saying about Wan 2.5 on X (Twitter)
Wan 2.5 / Kling 2.5 Turbo ... Esto no se detiene. Ya da un poco de miedo. El pangolín espectacular. Que opinas de mi caracol hecho con Kling 2.5 Turbo?
💥🤯 the new Wan 2.5 and it is insane now it has audio, sound effects and... voices! check these amazing examples below🧵
Wan 2.5 is so interesting. It solves the inverse problem of audio to video. In essence, it uses audio + prompt to infer video. This means you can design your audio track to tell a story which can be very efficient. Here I did a quick mock up first with drawing, then I fed in an Show more
With creative WAN product placement style that feel catchy + meme-worthy + launch-ready With @higgsfield_ai WAN 2.5 is here a fundamental update where every feature is powered by WAN. #HiggsfieldWAN
🚨 WAKE UP!! WAN 2.5 SPEAKS!🚨 You read that right, just released tonight, Wan 2.5 has native audio just like VEO3! Capable of 1080p and 10 seconds and Image To Video at launch. Text To Video Prompt: camera natural light, 8K. cinematic realistic dramatic zoom in on a a Show more
Wan 2.5 vs Kling 2.5 - the difference of physics is massive! 😱
Wan 2.5 Hits Real Off! I'm quite impressed how fluid the camera movement and sound design is. And of course with unlimited generation on Higgsfield AI I'm more than happy to experiment 🧩 Show more
What's Wan 2.5
Alibaba's advanced multimodal AI generation model with powerful text-to-video, image-to-video, and text-to-image capabilities
3 FormatsText/Image-to-Video & Text-to-Image
1080pMaximum Resolution
5-10 SecondsVideo Duration
MultilingualPrompt Understanding
Wan 2.5 is a cutting-edge multimodal AI model delivering versatile content generation across text-to-video, image-to-video, and text-to-image formats.
3 FormatsText/Image-to-Video & Text-to-Image
1080pMaximum Resolution
5-10 SecondsVideo Duration
MultilingualPrompt Understanding
What's Wan 2.5
Alibaba's advanced multimodal AI generation model with powerful text-to-video, image-to-video, and text-to-image capabilities
Wan 2.5 is a cutting-edge multimodal AI model delivering versatile content generation across text-to-video, image-to-video, and text-to-image formats.
Cutting-Edge Features of Wan 2.5
Explore the advanced multimodal capabilities that make Wan 2.5 stand out for all your image and video generation work
Multimodal Generation
Bakes text-to-video, image-to-video, and text-to-image generation into a single unified model, enabling smooth, uninterrupted creative workflows across all your different media types.
High-Resolution Output
Generates crisp videos up to 1080p resolution with flexible 480p and 720p output options, delivering professional-grade visual content for every type of use case.
Flexible Duration Control
Build videos with fully customizable runtimes from 5 to 10 seconds, adapting easily to fit all your unique content needs and creative requirements.
Audio Synchronization
Includes one-pass audio-video synchronization that supports custom audio integration and built-in automatic lip-sync for polished character animations.
Multiple Aspect Ratios
Works with both landscape (16:9) and portrait (9:16) formats across all supported resolutions, ideal for social media, presentations, and every common display format.
Multilingual Prompts
Processes prompts across multiple languages with native built-in translation support, opening the model up to global creators and diverse worldwide audiences.
Prompt Expansion
This advanced prompt optimization tool automatically enhances your input descriptions to produce far richer, more detailed generation results.
Negative Prompting
Refine your final outputs by calling out unwanted elements, putting precise control over your final generation quality and content directly in your hands.
Seed Control
Delivers fully reproducible results with customizable seed values, making it easy to get consistent outputs and iteratively refine your creative work.
Fast Generation Mode
Optimized speed-focused variants for both text-to-video and image-to-video tasks deliver matching high quality with dramatically reduced total processing time.
Custom Image Sizes
Text-to-image generation supports flexible dimensions from 256×256 to 1536×1536 pixels with multiple preset aspect ratios and fully custom sizing options.
Advanced Architecture
Built on Alibaba's cutting-edge video generation technology with sophisticated native understanding of motion, physics, and consistent visual coherence.
Frequently Asked Questions About Wan 2.5
Still have questions?
Developed by Alibaba, Wan 2.5 is an advanced multimodal AI generation model built with three core powerful capabilities: text-to-video, image-to-video, and text-to-image generation. Unlike single-purpose competing models, it delivers versatile performance across multiple content formats, with native support for 1080p resolution, flexible 5-10 second video durations, and built-in audio synchronization features.
Wan 2.5 supports multiple video resolutions including 480p (832×480), 720p (1280×720), and 1080p (1920×1080) in both landscape (16:9) and portrait (9:16) orientations. Video durations are flexible, with 5-second and 10-second options that let creators pick the right length for their specific project needs.
Wan 2.5 comes with advanced audio synchronization capabilities that let you integrate custom audio URLs directly into your video generation process. The model automatically aligns your audio with generated video content to create fully synchronized multimedia outputs, and accepts audio files in MP3, WAV, or M4A formats up to 50MB in size.
Wan 2.5 offers three primary generation modes: Text-to-Video creates dynamic videos from text prompts with customizable resolutions and durations; Image-to-Video transforms static images into animated videos; and Text-to-Image generates high-quality images with artistic capabilities and flexible aspect ratios from 256×256 to 1536×1536 pixels.
Yes, Wan 2.5 supports native multilingual prompt understanding. The model includes built-in translation options that convert prompts to English for optimal processing. It also features prompt expansion capabilities that enhance your input prompts for better generation results, making it accessible to creators worldwide.
Wan 2.5 offers two generation speed options for video creation. The standard mode provides a balanced mix of output quality and processing time, while the fast mode accelerates generation speed for quicker turnaround, which is ideal for rapid prototyping and iterative workflows. Both modes maintain high-quality output and support the same full set of resolution and duration options.
Wan 2.5's text-to-image mode supports multiple aspect ratios including 1:1 (1024×1024), 3:4, 4:3, and 16:9 formats, with high-definition options up to 1536×1536 pixels. The model features excellent prompt understanding, strong artistic capabilities, negative prompt support for avoiding unwanted elements, and custom ratio controls with dimensions ranging from 256 to 1536 pixels in 64-pixel increments.
Absolutely! Wan 2.5 supports both landscape (16:9) and portrait (9:16) aspect ratios across all supported resolutions. This built-in flexibility makes it perfect for a wide range of platforms and use cases, from traditional widescreen content to mobile-optimized vertical videos for social media platforms like TikTok and Instagram Reels.
Flexible AI Pricing
Pay-as-you-go credits or subscription plans. No hidden fees, cancel anytime.