LLM Stats

LLM Stats is an independent leaderboard and comparison site for AI models. It ranks more than 300 models from providers such as Anthropic, OpenAI, Google, Meta, Moonshot AI and Alibaba on a composite LLM Stats Score, built from public benchmarks and live API metrics and refreshed continuously. The main table shows each model's score alongside reasoning, coding and agent sub scores, context window, output speed, price per million tokens and license type, so open weights and proprietary models can be compared side by side. Separate leaderboards cover coding, writing, math, research, long context, tool calling, reasoning, image generation, video creation, text to speech, speech to text and embeddings. A pricing comparison page lists input, cached input and output prices per million tokens and an observed cost per turn, and it marks which models cost more or less than the median. The site also offers a data API and a tool called Superagent. It is aimed at developers, researchers and teams choosing which model to build on or buy access to. The about page describes a two person team that wants open and reproducible evaluation in place of benchmarks chosen by model vendors. The site does not publish plan prices for the leaderboard itself. The model prices it shows are the providers' own rates, not charges made by LLM Stats.