Hi everyone—Joon here from MVNT.
Welcome to our first newsletter, and thank you for joining us from the beginning.
mvntSTUDIO has now reached 2,000 users—nearly three times last month’s total—through organic growth alone. We’re also seeing our first paid creators and enterprise teams, with 5,500 monthly active users across the platform.
We’re developing models for specific dance genres, testing templates, and planning a dedicated mobile app. Here’s what you can try now and what we’re working on next. We’re still early, and we’re glad you’re here.
1. Better dance generation
On September, we’re releasing a major model upgrade. Rather than relying on a single model that handles every style adequately, we’re developing specialized models for individual genres. We’re starting with J-pop (kawaii-style) dance, with a focus on short-form dance challenges. The result is more detailed motion, improved hand and finger movement, and training supported by a much larger dataset. Watch the preview.
Behind the scenes, we operate our own motion-capture studio nearly every day. We record, clean, label, and review each batch of data—a hands-on process that gives us greater control over quality.
What’s tricky about buiding dance AI is that each dance genre has its own structure. K-pop, hip-hop, J-pop idol choreography, and TikTok challenges differ not only in appearance, but also in repetition, rhythm, and how movements land on the beat. We call this dance grammar. Our approach is to improve one genre at a time, with K-pop and hip-hop coming next.
2. Simpler creation experience
Creating dance digitally is still unfamiliar to many people. Today, a first-time user takes an average of 50 seconds to create a dance, so we’re working to make the process faster and more intuitive.
Our biggest upcoming feature is templates. Instead of starting with a blank prompt, you’ll be able to choose a style or structure and add your music. If you’ve used tools such as Higgsfield or Kling, the experience will feel familiar—but it is designed specifically around dance and music.
Formation dance is also in development, with support for up to three dancers at once. It’s still early, but this brings MVNT closer to the way choreography is often created and performed.
3. Mobile app is coming soon
About two-thirds of MVNT users access the product on mobile, so we’re now building a dedicated app. We’ve also learned that a monthly subscription does not suit everyone. Many users prefer one-time purchases, credits, or a flexible pay-as-you-go option.
We’re preparing a closed beta for the app. To request early access, complete the short survey below.
4. What we’re hearing from users
Common requests include improvements to the ComfyUI API, full-length dance generation, FBX import and export for custom models, and better foot contact. All are in development, and we’ll share updates as they become available.
We’re also improving integrations with tools such as Suno, Tripo, and Higgsfield, so MVNT can fit more naturally into existing workflows. Tell us what you want to create, which genres matter most, whether you’re interested in the beta, and what still feels confusing or incomplete. The survey takes about two minutes, and your feedback will help shape what we build next.
Selected beta testers will receive early access and free credits.
5. What’s next
What began as an experiment is becoming a useful creative tool for a wide range of people—not only dancers, but also humanoid robotics developers, AI researchers, VTuber and MMD creators, musicians, and cheerleading teams. Generating human dance from music has applications far beyond what we initially expected.
Next, we’re focusing on better models, mobile-first creation, stronger creator support, and community-led discovery. We’re also planning community sessions; use the survey to let us know if you’d like to participate.
Have a question, bug report, or idea? Contact us through Intercom on the MVNT website, or book a short call with Joon.
Thank you for being part of MVNT from the beginning. Start creating!
Best, Joon
P.S. MVNT was recently featured in Business Insider. If you cannot access the article, you can read the blog version. Thanks to Cheryl for the feature.
P.P.S. Two weeks ago, we presented MVNT at Unreal Fest Seoul 2026, with more than 60 people in the room and over 2,000 viewers on the livestream. A recap is coming soon.
