Overview

Arena Intelligence runs a crowdsourced platform that evaluates and compares large language models. It shows two anonymous models’ responses to the same prompt side-by-side, collects user votes, then reveals model identities; the resulting preference data feeds Elo ratings published on a public leaderboard. The platform serves researchers, developers, and AI enthusiasts who reference Arena’s rankings to gauge model progress, with the FastChat framework powering the interface and backend as open source. Its goal is to provide transparent, real-world evaluation and analytics to help AI labs and enterprises improve model performance while staying neutral and community-driven.

Launched Recently

Significant Headcount Growth