BrowseComp

What it measures

OpenAI's benchmark for browsing agents that must find hard-to-locate, short-answer facts on the web.

Primary metric

short-answer accuracy (%)

Judge model

not judged by a model

Human baseline

29.2OpenAI's verification campaign reports trainers solved 29.2% of 1,255 checked problems within the allowed search process; among solved problems, trainer and reference answers matched 86.4%.

Languages

English

Health

BrowseComp uses short answers to simplify grading. Scores depend strongly on browser/search tool quality, time budget, and whether the agent is allowed repeated attempts or parallel search.

Status evidence

BrowseComp is nearing saturation: launch baseline Deep Research 51.5% (Apr 2025, openai.com/index/browsecomp); by Aug 2026 SOTA reached 92.2% (GPT-5.6 Sol) with top-3 within 1.4 pts (llm-stats.com/benchmarks/browsecomp, benchlm.ai) and only ~7.8 pts headroom. LoHoSearch reports 'rapidly saturated... surpassing 90%... 30%->90% in barely ten months' (arxiv.org/abs/2606.12837) and Anthropic shows eval-awareness/decrypted answer key (anthropic.com/engineering/eval-awareness-browsecomp); LiveBrowseComp confirms IKD leakage (arxiv.org/abs/2605.28721).

Saturated since

not saturated

Refresh cycle

static

Test set public

yes

OpenAI asks users not to reveal examples online and includes a canary string for training-data filtering, but the benchmark is public and static.

Models scored on this benchmark

The reverse lookup, which is this site’s addition: every model here with a score on this benchmark. Independent measurements and vendor-reported numbers are kept apart rather than interleaved, because they are different claims.

No model on this site has a score on this benchmark yet. That usually means the models that were measured on it are not in the registry, rather than that nobody has run it.

Trajectory