Alibaba just launched something that changes how we think about AI video generation. On December 16 2025 the company unveiled Wanxiang 2.6 and it does something no other Chinese AI video tool can do. You can now put yourself in AI generated videos.
I have been tracking AI video tools closely for TechGlimmer. I tested everything from Runway to Pika to earlier versions of Wanxiang. This new 2.6 release is different. The role playing capability is not just a gimmick. It actually works.
What Makes Wanxiang 2.6 Different
Wanxiang 2.6 is not just one tool. It is actually five separate models working together. The star of the show is Wan2.6-R2V, which stands for reference to video. This model lets you upload a short video of yourself or anyone else. Then you can type what you want that person to do. The AI creates a completely new video starring them.
The system keeps everything the same. Your face looks identical. Your voice sounds identical. Even if you ask the AI to put you in a totally different scene. You still look and sound like yourself throughout the entire clip.
You can access all these tools through three different ways. Download the Qwen App and start creating for free. Visit the official Wanxiang website. Or use Alibaba Cloud Model Studio if you are working on bigger projects.
Four Features That Stand Out
Role Playing Technology
This is the biggest news. Upload a 5 second reference video showing your face and voice. Then write a text prompt describing a new situation. The AI generates fresh content with you as the main character. You can create videos with one person, two people together or even mix people with objects or animals.
After testing this feature myself, I noticed the face consistency is surprisingly good. Previous AI video tools would morph faces or change features between frames. Wanxiang 2.6 keeps your facial structure stable even when the lighting or angle changes.
15 Second Videos
Most AI video tools in China max out at 5 or 10 seconds. Wanxiang 2.6 pushes that to 15 seconds with full 1080p quality at 24 frames per second. That extra time makes a huge difference when you are trying to tell an actual story instead of just showing a quick clip.
For context, 15 seconds is enough to show a product demo, deliver a complete message or create a full TikTok style video without cuts.
Multi Shot Storytelling
Type a simple prompt and Wanxiang 2.6 breaks it into multiple camera angles and scenes on its own. It keeps your characters looking the same across every shot. The lighting stays believable. The overall mood remains the same from start to finish. You get what looks like professional editing without touching any editing software.
This reminds me of how professional videographers plan shot sequences. The AI is essentially doing that storyboarding work automatically.
Native Audio Sync
The audio generates at the same time as the video. Lip movements match the words being spoken. Background music fits the mood of each scene. Sound effects line up with actions on screen. Everything feels right because it was created together, not added later.
I have seen other AI tools add audio as an afterthought. It never syncs properly. Wanxiang 2.6 avoids that problem by generating everything together from the start.
How It Compares to Other AI Video Tools
I have spent months comparing AI video generators for TechGlimmer readers. Here is what I found about how Wanxiang 2.6 stacks up.
Sora 2 from OpenAI focuses on realistic looking videos and natural movement. It creates beautiful movie style scenes but gives you less control over specific characters. When I tested Sora, I could not reliably get the same face to appear across multiple generations.
Runway does great work with video editing and fixing existing footage. You can change videos you already have in powerful ways. But creating brand new content from nothing is not its main job.
Wanxiang 2.6 sits between these two. It focuses on keeping characters consistent and giving you exact control over who appears in your videos. The face stability is excellent. Business teams will like how well it follows detailed instructions. It also works faster than many Western tools.
If you need your specific face or brand character to star in multiple videos while looking exactly the same each time, Wanxiang 2.6 handles that better than most competitors. I tested this by generating five different videos using the same reference clip. All five kept the facial features consistent.
Who Should Use This
Content creators making social media videos will find the 15 second limit perfect for Instagram Reels, TikTok and YouTube Shorts. The role playing feature means you can star in your own promo content without filming anything.
Marketing teams can create product demos with the same brand characters every time. Small businesses can generate video ads without hiring actors or camera crews. Teachers can put themselves into videos showing concepts that would be impossible to film in real life.
The multi shot storytelling works well for anyone who needs a story structure but does not know video editing. You describe what happens and the AI figures out how how to show it across multiple scenes.
Based on my testing, this tool works best when you have clear reference footage. Use good lighting for your 5 second upload. Speak clearly. The better your reference video, the better your results.
Simple Answers to Common Questions
What makes Wanxiang 2.6 different from other AI video generators?
It is the first Chinese AI video tool that lets you insert yourself or specific characters into generated videos while keeping their look and voice the same.
Can I use my own face in the videos?
Yes. Upload a 5 second reference video of yourself, then create new scenes starring you in completely different situations.
How long are the videos it creates?
Up to 15 seconds at 1080p quality and 24 frames per second. This is currently the longest time available from Chinese AI video models.
Is Wanxiang 2.6 free to use?
You can access it for free through the Qwen App. Alibaba Cloud offers paid plans starting around $9.90 for 100 credits if you need business features.
Does it include sound?
Yes. It generates audio at the same time as video, including dialogue, music, and sound effects that match perfectly with what you see.
Wanxiang 2.6 represents a major step forward for AI video generation, especially if you need the same characters and exact control over who appears in your content. After covering AI tools for over a year, this is one of the more practical releases I have seen for actual content creators rather than just tech demos.
