Arena AI leaderboard hits $100 million revenue run rate eight months after launch

Arena, the crowdsourced AI model leaderboard that began as a UC Berkeley research project, said on June 29, 2026 it had crossed a $100 million annualized revenue run rate, eight months after launching its commercial AI Evaluations service in September 2025. The milestone, announced by Arena, tracks a fast turn from open research tool to paid business.

One caveat sits at the center of the number. Arena’s revenue is consumption-based, billed per model-evaluation run, not recurring subscription revenue, and the company itself flags that the $100 million is a run rate rather than traditional annual recurring revenue. The figure is Arena’s own and has not been independently audited.

The reach behind the business is the harder thing to replicate. The platform has logged more than 700 million conversations and 82 million votes and draws over 10 million monthly visitors, Arena said, the crowd-sourced comparisons that produce its widely cited model rankings. Its newer Agent Mode records more than 5 million turns a month, growing about 10% week over week.

Arena incorporated in April 2025 and rebranded from LMArena in January 2026. It is led by co-founders Anastasios Angelopoulos as chief executive, Wei-Lin Chiang as chief technology officer, and UC Berkeley professor and Databricks co-founder Ion Stoica. The company has raised $250 million total, including a $150 million Series A in January 2026 at a $1.7 billion post-money valuation, when its run rate was $30 million.

That means the run rate has roughly tripled since the raise, on the company’s telling. Consumption revenue can also contract as fast as it grows if a few large customers cut evaluation spend, which the milestone post does not address.