Which One is Better? Vote Now and Reveal the Model Used
Total votes include lma Studio Arena and Discord participations
Got a Video Prompt Idea? Join the Battle Ground
and Add it to the Evaluation Pool!
All video generations' tasks are created in Discord Channel. Join our server to enter prompts and get a series of 6 videos with one prompt as professional creator.
Spotlight Channel:

Leaderboard
The following leaderboard shows the comparison of human evaluation with AI. In collaboration with
Nanyang Institute, we adopt one of the most popular video model assessment frameworks in the industry.
VBench is the world's first benchmark to cover 16 dimensions of video generation, recognized as a CVPR 2024 Highlight. It not only breaks down "video generation quality" with fine-grained precision but also pioneers human preference alignment, ensuring objective metrics match human perception. With VBench++, it further supports both text-to-video and image-to-video tasks, while evaluating not just technical quality but also model trustworthiness—establishing itself as the authoritative standard and compass for the industry. Github
Text-to-Video Arena
Compare models according to their ability to generate videos based on the given prompt. Generate videos and vote in the Discord server. Discord server
Rank ELO | Model | Human Votes ↕️ | ELO ↕️ | 95% CI | Organization | License |
---|---|---|---|---|---|---|
No model data available |
Why Ima Studio Arena?
Video performance isn't just about generating content—it's about understanding what resonates. Ima Arena evaluates every video through:
- Human preference: Community votes capture taste and creativity
- Model scoring: AI provides consistency and precision
This creates benchmarks that are both authentic and scalable.
What you gain:
- Visibility – Videos enter structured evaluations beyond social feeds
- Recognition – Votes and rankings build measurable reputation
- Rewards – Arena participation unlocks
Why we're building this:
Ima Studio empowers creators through AI tools. We believe evaluation completes the creative loop—helping creators understand performance while building transparent systems that benefit both creators and AI development.