Search Arena
A crowdsourced human-preference leaderboard from LMArena that ranks search-augmented LLMs via blind pairwise votes on grounded, web-search answers, reported as Bradley-Terry Elo-scale ratings.
What this benchmark measures
A crowdsourced human-preference leaderboard from LMArena that ranks search-augmented LLMs via blind pairwise votes on grounded, web-search answers, reported as Bradley-Terry Elo-scale ratings.
Rows on this page are sourced from public benchmark artifacts, leaderboard exports, or source-linked model reports. Each row keeps benchmark version, source model name, and available run details attached to the score.
The metric shown here is Elo. It should be interpreted within Search Arena, not compared as part of a site-wide ranking.
Frequently asked
What is Search Arena?
A crowdsourced human-preference leaderboard from LMArena that ranks search-augmented LLMs via blind pairwise votes on grounded, web-search answers, reported as Bradley-Terry Elo-scale ratings. It is a chat preference benchmark measured by Elo.
What does Elo mean on Search Arena?
Search Arena reports Elo; higher is better. Scores are shown only within Search Arena and are never averaged with other benchmarks.
What is the top reported Search Arena score?
Claude Opus 4.6 has the top reported score on Search Arena: 1251 (Elo).
Why do Search Arena scores differ across runs?
Harness, scaffold, reasoning effort, and prompt setup change results, so two runs of the same model can differ. evals.report keeps each score with its run context so the differences stay visible.
Does evals.report rank models across benchmarks?
No. Search Arena scores are shown within their own metric; evals.report never combines benchmarks into a composite ranking or a single "best model".