The leaderboard “you can’t game,” funded by the companies it ranks

Leaderboard You Can’t Game, Backed by Companies It Ranks

Arena has become the go-to AI model ranking platform, evolving from a research project to a $1.7 billion startup in seven months

Technology

San Francisco: Arena is a special website that ranks the best artificial intelligence models people can use today. It started as a research project at UC Berkeley and now is worth $1.7 billion. People call it the most trusted place to see which AI model is the smartest.

The founders, Anastasios Angelopoulos and Wei-Lin Chiang, explain that their website works differently from other tests. It’s harder to trick because real people use the AI models and give feedback. Big tech companies like OpenAI, Google, and Anthropic give money to support Arena, but it tries to stay fair and not favor any company.

Right now, a model called Claude is getting the best scores in special tests for legal and medical work. Arena wants to test more than just chat – they want to check how well AI models can write code and do real jobs. They made a new product for businesses that want to test their own AI models.

Image Credits and Reference: https://techcrunch.com/video/the-leaderboard-you-cant-game-funded-by-the-companies-it-ranks/