BenchLeader: LLM ranks where you browse

See where a model ranks on independent benchmarks, what a task costs and a better-value alternative, on OpenRouter and Hugging Face.

Know how good a model really is before you pick it. BenchLeader adds a small card to model pages on OpenRouter and Hugging Face. It shows where the model ranks on BenchLeader's daily leaderboard, which combines results from independent public benchmarks (Artificial Analysis, LMArena, Epoch AI, SWE-bench, Vals AI and more) rather than vendor launch claims. On each model page you see: - Its rank overall and in coding - Its BenchLeader Index score - What one benchmark task costs on it, or its price per million tokens - Its output speed in tokens per second - A better-value alternative when one exists: a model that ranks higher for the same price, or one nearly as good for less than half Click the toolbar icon to look up any model by name. Every number links to the full results on benchleader.com, where each score links back to the benchmark that published it. Data updates daily. Privacy: the extension only reads the address of the tab you are on, and only on openrouter.ai and huggingface.co. When that page is a model page, it sends the model's name to benchleader.com to look up its rank. Nothing else is collected or stored: no browsing history, no page content, no account. Full policy: https://www.benchleader.com/privacy#browser-extension Free, no sign-up.