Arena (formerly LMArena) lets you chat with multiple cutting-edge AI models, see their answers side by side, and choose which response you prefer. Your prompts and votes are used (with care for privacy and transparency) to fairly evaluate models and improve how they work for real people. Over 5 million community members, including builders, researchers, and creative professionals, use Arena both to get work done and to shape how future AI systems are designed.
Key Features
Community voting system (“battle mode”) where you compare two anonymous model responses and choose the better one, powering public leaderboards.
Unified chat interface to chat with AI, build apps and websites, write and edit text, search the web, and generate images and videos.
Comprehensive leaderboards covering overall performance and specific tasks like text, code, text-to-image, image edit, text-to-video, image-to-video, vision, and search.
Open research program with free public datasets of human preferences and multiple peer-reviewed research papers on AI evaluation and human feedback.
Use Cases
Chatting with AI to write, edit, search, or generate images and videos while automatically contributing feedback to model evaluations.
Comparing cutting-edge models side by side to decide which one to use for coding, content creation, or product integration.
Exploring public leaderboards and research datasets to study model quality, safety behavior, and user preferences for academic or industry work.