Top models, ranked.

Token volume, benchmark scores and prices. All sortable, all in one table.

How it works? Every night new data is fetched and embedded into this page. The browser makes no extra requests and needs no API key. Token volume is the sum of daily rankings-daily records over the embedded range. Scores are matched from artificial-analysis (Coding, Intelligence, Agentic), design-arena (ELO) and built-in benchmarks (GPQA, Tau-bench, search). The apps tab is an app-rankings snapshot (4 groups, 15 subcategories), tasks are a classifications snapshot (Code, Data, Agent, General). Rankings slugs carry versions (...-2024-05-13, :free), score slugs are bare (openai/gpt-4o). Exact match is tried first, then prefix match. Cells without a match stay empty. Click a row to open every field of the raw record.
LLMRank.xyz, CC BY 4.0 attributed data