AI evaluation platform Arena has raised $200m in a Series B funding round, valuing the business at $3.1bn just 10 months after its previous investment round.
The funding was led by Lightspeed Venture Partners and Khosla Ventures, with participation from Salesforce Ventures, 01 Advisors, Dell Technologies Capital, Endeavor Catalyst, Andreessen Horowitz (a16z), Felicis and other investors.
Founded in 2023 as a research project at the University of California in Berkeley, Arena developed a crowdsourced platform that allows users to compare the performance of different artificial intelligence models. Users submit prompts to two models and select which produces the better response, with the results contributing to a public leaderboard ranking their relative performance.
The platform has attracted tens of millions of monthly visitors, providing AI developers with feedback on how their models perform across a range of real-world tasks.
Arena announced a $150m Series A in January at a post-money valuation of $1.7bn. At the time, the company reported annualised revenue of $30m, before reaching $100m in annualised run-rate revenue in June. The latest funding round brings its valuation to nearly double that level in less than a year.
The business began generating revenue in September 2025, when it introduced AI Evaluations, a commercial service providing AI model developers and enterprise customers with detailed performance analytics based on feedback from its user community.
The launch came as AI companies faced growing questions over the reliability of conventional benchmarking methods. Developers have increasingly sought ways to assess how models perform outside controlled testing environments, while businesses looking to adopt AI have needed clearer evidence of which systems best suit their requirements.
Arena’s community-driven approach provides an additional source of performance data, allowing companies to assess models using feedback from people interacting with them directly.
The company has now expanded its leaderboard to include an alignment category, designed to assess how models behave when responding to human instructions. The rankings examine issues including unauthorised actions, incorrect attribution of information and cases where a model claims to have completed a task when it has not.
Arena said in its funding announcement that the rapid development of AI has made independent evaluation increasingly important, particularly as models become more capable and conventional benchmarks become less reliable.
The company is positioning its platform as a way to measure both model performance and behaviour in practical settings, giving developers and businesses additional information when deciding which systems to use.
Its preliminary alignment leaderboard currently places several OpenAI models at the top, while Anthropic’s Claude Opus 5.5 and Claude Fable rank sixth and ninth respectively.
The new funding will support Arena’s efforts to develop its AI evaluation platform and expand its work measuring how models perform in real-world applications. The company is also seeking to establish its alignment assessments as a further measure of whether AI systems can be trusted to act in line with human intentions.
Arena’s latest round comes as demand grows for tools that help businesses compare an expanding range of AI models, with the company building a commercial business around a platform that remains free for consumers to use.
