🧠 Food for thought

Implications, context, and why it matters.

LMArena hit a $30 million annualized run rate in under four months

• LMArena runs a platform for evaluating AI models and launched a paid AI Evaluations service in September 2025, hitting a $30 million annualized consumption run rate by December 2025, which it calls Annual Recurring Revenue (ARR) 1.
• Years of free community work set the stage 1. More than 5 million monthly users across 150 countries create over 60 million conversations each month, which became a pool of evaluators at launch 1.
• Clients include AI labs and enterprises in software engineering, law, medicine, plus scientific research that pay for evaluations, which turned that community into a business 1.

Developers can build AI governance tools on LMArena’s open evaluation stack

• A third-party FAQ says LMArena offers an API that supports FastChat (an open-source chat protocol for large language models) plus OpenAI compatible formats, which gives developers integration points for compliance and reporting tools built on its evaluation data 2.
• EU AI Act rules require documented model testing, while the National Institute of Standards and Technology (NIST) AI Risk Management Framework calls for ongoing performance checks. Plugins could auto generate compliance reports from LMArena evaluation results.
• Developers could ship Machine Learning Operations (MLOps) connectors that plug LMArena evaluations into Continuous Integration/Continuous Delivery (CI/CD) pipelines 1. They could also build dashboards for non-technical stakeholders to read model results across coding, reasoning, or professional domains 1.