> ## Documentation Index
> Fetch the complete documentation index at: https://aegean.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Where Frontier Models Are Ranked

> Leaderboards that compare frontier AI models.

A model developer's own benchmark table reports results on the tests they chose, run with the prompts and settings they chose. The leaderboards below are run by third parties instead. Each one measures something different, so a model can lead one board and sit lower on another.

> [card] [Artificial Analysis](https://artificialanalysis.ai/leaderboards/models)
>
> Runs its own tests on each model it lists, with one fixed method, and combines them into the Intelligence Index. It also reports speed, latency and price for each API provider.

> [card] [Epoch AI Benchmarking Hub](https://epoch.ai/benchmarks)
>
> Runs its own evaluations, including GPQA Diamond, FrontierMath and SWE-bench Verified. You can inspect the prompt, response and score for every question, and download the data.

> [card] [Arena](https://arena.ai/leaderboard)
>
> Ranks models by Elo from blind side-by-side votes, with separate boards for text, coding, web development and vision. Formerly Chatbot Arena and LMArena.

> [card] [Scale SEAL leaderboards](https://labs.scale.com/leaderboard)
>
> Vetted domain experts grade models on private datasets, so developers cannot train on the test prompts. Each capability has its own leaderboard.

Arena measures which answer people prefer. That preference rewards length and formatting as well as correctness, so Arena offers a style control filter to reduce the effect. The other three grade answers against a reference or an expert rubric.

***

Found an error on this page? [Report an issue](https://github.com/aegean-ai/eaia-feedback/issues/new?template=page-issue.yml\&page=/research/model-benchmarks).


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.