Wan 2.7 AI Video GeneratorText to Video & Image to Video
Alibaba's most advanced AI video model. Subject reference, voice reference, multi-image grid to video, instruction editing — all available on Topview Board.
[Duration]: 15 seconds [Camera]: Sony A7S III [Style & Type]: Middle East business consulting advertisement. Realistic style, high-brightness office lighting, using panning shots and fast multi-scene switching, professional and atmospheric. [Golden 3 Seconds]: A Western man in a suit sitting at a desk piled with documents, opens a notebook and looks at the camera, with huge subtitles appearing on screen. [Video Content]: The man wears black-rimmed glasses with a confident and professional expression. Actions include flipping through pages, walking in the office area @Image16, explaining with open hands, and walking confidently towards the camera at the end, finally popping up the brand CTA @Image17 [Pace & Atmosphere]: Fast-paced editing with dynamic background music. Smooth transitions, using large golden keywords to enhance visual impact, strong business atmosphere. [Dialogue]: For a long time we have been helping people setup their businesses here in the UAE. And now, it's time for us to help you with your numbers as well... we don't just set up your company, we keep it compliant and financially healthy.
What Types of Videos Can You Make with Wan 2.7?
Wan 2.7 is not just for one format. It works especially well for video types that need stronger motion, cleaner scene consistency, more cinematic framing, or reference-driven storytelling.
Action & Motion-Heavy Videos
Use Wan 2.7 for fast-moving scenes where body motion, momentum, and timing need to stay readable. This works well for sports-style edits, cinematic action sequences, game trailers, and product demos with movement.
"Cinematic action sequence with a female soldier in full tactical gear moving through snowy industrial ruins. Emphasize fast body motion, believable weapon recoil, airborne debris, cold atmospheric lighting, and a tense battlefield rhythm with clean shot-to-shot continuity."
Cinematic Character Shots
Wan 2.7 is a strong fit for dramatic portraits, fashion-style visuals, moody trailers, and music-video moments where framing, depth, and directional lighting need to feel more intentional and less flat.
"Moody cinematic portrait of a woman with a black bob haircut and pearl necklace walking down a dim tiled staircase at night. Use a smooth backward tracking shot, serious expression, soft directional light, deep shadow contrast, and elegant framing with strong visual depth."
Product Story & Commercial Videos
Wan 2.7 can turn products into more premium-looking visual stories. Use it for ad creatives, launch teasers, beauty shots, product reveals, and commercial sequences that need controlled camera motion and cleaner visual polish.
"Premium product commercial featuring a luxury skincare bottle on a reflective pedestal. Use a slow cinematic push-in, soft specular highlights, precise shadow control, glossy studio reflections, and an elegant reveal rhythm that feels polished, minimal, and high-end."
Reference-Driven Story Videos
Because Wan 2.7 supports subject reference, voice reference, and multi-image input, it is useful for creator-style storytelling, talking scenes, avatar-led explainers, and consistent multi-shot character narratives.
"Reference-driven creator video of a young speaker talking directly to camera in a warm indoor studio. Keep facial identity consistent across cuts, preserve natural hand gestures and clean lip sync, and use soft cinematic lighting with documentary-style framing for an authentic storytelling feel."
What Is Wan 2.7?
Wan 2.7 is Alibaba's latest AI video generation model, scheduled for March 2026. Built on Diffusion Transformer (DiT) and Mixture of Experts (MoE) architecture with 27B+ parameters, it generates 1080p videos up to 15 seconds from text, images, or multi-image grids — with native lip-sync and subject consistency. The model uses a Causal 3D VAE with 4x8x8 compression ratio, activating only 14B of 27B total parameters per token to cut computation by 50% while improving complex scene generation.
Diffusion Transformer (DiT)
Advanced transformer-based diffusion model enabling superior temporal coherence and complex motion understanding.
Mixture of Experts (MoE)
27B total parameters with 14B activation, reducing computation by 50% while improving multi-character interactions.
Causal 3D VAE
Efficient spatiotemporal compression with 4x8x8 ratio, supporting arbitrary-length 1080p video encoding.
What's New in Wan 2.7
Six breakthrough upgrades over Wan 2.6 — from subject consistency to instruction-based editing.
Subject Reference
Upload a character reference to maintain visual identity across generated scenes.
Voice Reference
Provide a voice sample and the model preserves that vocal style in generated audio.
Multi-Image Grid to Video
Feed a 9-grid layout of images and Wan 2.7 synthesizes them into a coherent video sequence.
Instruction Editing
Edit generated videos with text instructions — change scenes, adjust pacing, swap elements without regenerating.
Video Replication
Upload a reference video and recreate its style, pacing, and composition with new content.
Quality Full Upgrade
Across-the-board improvements in visual quality, audio fidelity, motion dynamics, style consistency, and character coherence.
Wan 2.6 vs Wan 2.7
| Feature | Wan 2.6 | Wan 2.7 |
|---|---|---|
| Subject Reference | Limited | Full support |
| Voice Reference | No | Yes |
| Multi-Image Grid | No | 9-grid to video |
| Instruction Editing | No | Yes |
| Video Replication | No | Yes |
| Video Reference Clips | 1 | Up to 5 |
| Real Person Input | Limited | Full support |
| Quality / Audio / Dynamics | Baseline | Full upgrade |
How to Use Wan 2.7 in Topview (3 Steps)
Enter a prompt
Describe the video you want using natural language.
Generate Video
Click generate and watch Wan 2.7 bring your ideas to life in seconds.
Download the video
Export a clean MP4 when you're ready.
Wan 2.7 Core Capabilities
Everything you need to create professional AI videos — from text prompts to multi-reference generation.
Text to Video
Generate cinematic videos from text descriptions in English, Chinese, Japanese, Korean, and German.
Image to Video
Animate static images with natural motion, camera movement, and cinematic effects.
Subject + Voice Reference
Maintain character identity and vocal style across generated clips for brand consistency.
Multi-Image Grid to Video
Convert 9-grid image layouts into coherent video sequences with smooth transitions.
Instruction Editing
Modify generated videos via text commands — change scenes, adjust pacing, swap elements.
Native Lip Sync
Industry-leading audio-visual synchronization for talking-head and dialogue content.
Wan Model Evolution: From 2.1 to 2.7
Five generations of continuous innovation in AI video generation from Alibaba's Tongyi Lab.
Wan 2.1
Open source (Apache 2.0), consumer GPU (6GB+), 14B params, VBench 86.22%
Wan 2.2
MoE (27B total), 50% compute savings, character replacement, 60+ cinematic params
Wan 2.5
Photo singing/dancing, 10s generation, audio-visual sync, native multimodal
Wan 2.6
Full lip-sync, voice cloning, multi-shot narrative, 15s generation
Wan 2.7
Subject+voice reference, instruction editing, multi-image grid, quality full upgrade
Technical Specifications Comparison
| Metric | Wan 2.1 | Wan 2.2 | Wan 2.5 | Wan 2.6 | Wan 2.7 |
|---|---|---|---|---|---|
| Max Resolution | 720p | 720p | 1080p | 1080p | 1080p |
| Max Duration | 5s | 5s | 10s | 15s | 15s |
| Frame Rate | 24fps | 24fps | 24fps | 24fps | 24fps |
| Parameters | 14B | 27B | 27B+ | 27B+ | 27B+ |
| VBench Score | 86.22% | 87.5% | — | 89%+ | 89%+ |
Wan 2.7 vs Other AI Video Models
See how Wan 2.7 compares to the leading AI video generation models available today.
| Metric | Wan 2.7Recommended | SeedDance 2.0 | Sora 2 | Kling 3.0 | Veo 3.2 | Gen-4.5 |
|---|---|---|---|---|---|---|
| Max Duration | 15s | 15s | 25s | 25s | 10s | 10s |
| Resolution | 1080p | 1080p | 1080p | 4K/60fps | 1080p | 1080p |
| Open Source | Yes (2.1) | No | No | No | No | No |
| Real Person Input | Yes | No | — | — | — | — |
| Video Reference Clips | 5 | 1 | — | 1 | — | 1 |
| Instruction Editing | Yes | No | — | — | — | — |
| Lip Sync Quality | ★★★★★ | ★★★★ | ★★★★ | ★★★★ | ★★★★ | ★★★ |
Why Use Wan 2.7 on Topview
Access Wan 2.7 alongside every leading AI video model — in one workspace, one subscription.
All-in-One Board
Access Wan 2.7 alongside Sora, Veo, Kling, Runway, and more on one canvas. No need to switch between platforms.
Real-Time Collaboration
Share your Board with teammates. Comment, annotate, and iterate on AI video outputs together in real time.
Single Subscription
One Topview plan covers all integrated models. No per-model pricing or separate API keys required.
Marketing Video Workflow
Topview's AI Video Agent combines Wan 2.7 with product URL extraction, viral template matching, and UGC generation.
Export Flexibility
Download as MP4, MOV, or WebM. Choose resolution and aspect ratio for TikTok, Reels, Shorts, or paid ads.
All-in-One Creation Workflow
From image to video to publishing, Topview lets you complete the whole workflow in one place instead of switching between separate tools.
Start Free — Try Wan 2.7 on Topview Board
Experience Wan 2.7's subject reference, voice reference, and multi-image grid to video on Topview Board. No credit card required.
All AI video models in one workspace · Real-time collaboration · No credit card required