Arena

Arena is a public platform for evaluating AI models through real use. Visitors chat with models, compare anonymous responses side by side in battle mode, and vote for the better answer; those votes feed public leaderboards that rank frontier models by how people actually judge them. It covers text, code and web development, vision, image generation, video and, more recently, agents. It is useful for developers choosing a model for a coding or product workload, researchers tracking the frontier, and anyone who wants to try several leading models without separate subscriptions. The home screen offers build prompts such as a landing page, a dashboard, a browser game, design to code from an uploaded image, a full stack app or an online shop, so users can compare how different models handle the same software task. Leaderboards are broken out by category, including best coding agents, best WebDev models and best agents for work, and an agent mode lets users watch and judge agent sessions. The site notes that inputs are processed by third party AI and responses may be inaccurate. Arena is free to use on the web. It publishes no pricing page and no paid plans for users. Arena was created by researchers from UC Berkeley and was previously known as Chatbot Arena and then LMArena. In 2025 the team formed Arena Intelligence Inc., and the company reports $250 million in venture funding led by Felicis and UC Investments.