Google's platform for crowdsourced AI evaluation, with benchmarks, competitions, and Arenas where you vote on which model made the better output.
About Kaggle
Kaggle, at kaggle.com, describes itself as the leading platform for independent, crowdsourced AI evaluation. It runs competitions, benchmarks and hackathons, and hosts public models, datasets and notebooks. Kaggle is owned by Google.
Kaggle Arenas, launched in November 2026, are head-to-head matchups between AI models. Each matchup shows two anonymous outputs for the same prompt, side by side, whether that's a video, a web page, an image or text. You vote A is better, B is better, Both are good or Both are bad, and the votes update the leaderboard.
The Arenas page listed 15 arenas when checked on November 13, 2026, filtered by Video, Image, Web page and Text. HeyGen's Code2Video Arena tests models on prompt adherence when generating HyperFrames code. MineBench Arena covers 3D Minecraft voxel generation across 27 prompts. Kaggle's own arenas include Pelican Arena (an SVG of a pelican riding a bicycle), Chart Arena (a small dataset and a brief turned into one self-contained SVG chart), Floorplan Arena (models furnish hand-drawn room sketches) and Landing Page Arena (a one-page website for a small business from a client brief), along with ASCII, SVG, Joke, ELI5, Opening Lines, Corp Speak, Landmark, Explode and Sort Viz arenas. Several arenas state a pass rule an output has to meet before people rate it.
Each arena card shows the top-ranked model, its score, the output type and the number of votes so far.
Features
- Arenas show two anonymous outputs for the same prompt, side by side
- Vote A is better, B is better, Both are good or Both are bad
- Votes update each arena's leaderboard
- Arenas for video, image, web page and text output
- Arenas published by Kaggle and by outside teams such as HeyGen and MineBench
- Benchmarks from labs, researchers and companies, with model rankings
- Competitions and hackathons with prizes
- Public models, datasets and notebooks
FAQs
Common questions about Kaggle





