reasoning benchmark
BrowseComp Leaderboard
BrowseComp is a reasoning benchmark comprising 1,266 questions that challenge AI agents to navigate the internet to find hard-to-find, entangled information.
Updated Sep 5, 2026
Models63
Model coverage63
MetricScore
EvidenceB
BrowseComp Ranking
Higher score ranks better on this benchmark.
Sort by
Rank | Model | Score | Percentile | Participants | Evidence | Evaluated |
|---|
| Rank01 | ModelOP | Score91.50% | Percentile100.00% | Participants63 | EvidenceC | Evaluated |
| Rank02 | ModelMA | Score91.20% | Percentile98.39% | Participants63 | EvidenceC | Evaluated |
| Rank03 | ModelAN | Score90.80% | Percentile96.77% | Participants63 | EvidenceC | Evaluated |
| Rank04 | ModelOP | Score90.40% | Percentile95.16% | Participants63 | EvidenceC | Evaluated |
| Rank05 | ModelOP | Score90.10% | Percentile93.55% | Participants63 | EvidenceC | Evaluated |
| Rank06 | ModelOP | Score87.50% | Percentile91.94% | Participants63 | EvidenceC | Evaluated |
| Rank07 | ModelAN | Score86.90% | Percentile90.32% | Participants63 | EvidenceC | Evaluated |
| Rank08 | ModelMA | Score86.30% | Percentile88.71% | Participants63 | EvidenceC | Evaluated |
| Rank09 | ModelBY | Score86.20% | Percentile87.10% | Participants63 | EvidenceC | Evaluated |
| Rank10 | ModelGO | Score85.90% | Percentile85.48% | Participants63 | EvidenceC | Evaluated |
| Rank11 | ModelBY | Score84.90% | Percentile83.87% | Participants63 | EvidenceC | Evaluated |
| Rank12 | ModelAN | Score84.70% | Percentile82.26% | Participants63 | EvidenceC | Evaluated |
| Rank13 | ModelOP | Score84.40% | Percentile80.65% | Participants63 | EvidenceC | Evaluated |
| Rank14 | ModelAN | Score84.30% | Percentile79.03% | Participants63 | EvidenceC | Evaluated |
| Rank15 | ModelTE | Score84.20% | Percentile77.42% | Participants63 | EvidenceC | Evaluated |
| Rank16 | ModelAN | Score84.00% | Percentile75.81% | Participants63 | EvidenceC | Evaluated |
| Rank17 | ModelMI | Score83.52% | Percentile74.19% | Participants63 | EvidenceC | Evaluated |
| Rank18 | ModelDE | Score83.40% | Percentile72.58% | Participants63 | EvidenceC | Evaluated |
| Rank19 | ModelOP | Score83.30% | Percentile70.97% | Participants63 | EvidenceC | Evaluated |
| Rank20 | ModelOP | Score82.70% | Percentile69.35% | Participants63 | EvidenceC | Evaluated |
| Rank21 | ModelAN | Score79.30% | Percentile67.74% | Participants63 | EvidenceC | Evaluated |
| Rank22 | ModelZA | Score79.30% | Percentile66.13% | Participants63 | EvidenceC | Evaluated |
| Rank23 | ModelOP | Score77.90% | Percentile64.52% | Participants63 | EvidenceC | Evaluated |
| Rank24 | ModelTM | Score77.40% | Percentile62.90% | Participants63 | EvidenceC | Evaluated |
| Rank25 | ModelBY | Score77.30% | Percentile61.29% | Participants63 | EvidenceC | Evaluated |
| Rank26 | ModelMI | Score76.30% | Percentile59.68% | Participants63 | EvidenceC | Evaluated |
| Rank27 | ModelZA | Score75.90% | Percentile58.06% | Participants63 | EvidenceC | Evaluated |
| Rank28 | ModelMA | Score74.90% | Percentile56.45% | Participants63 | EvidenceC | Evaluated |
| Rank29 | ModelAN | Score74.70% | Percentile54.84% | Participants63 | EvidenceC | Evaluated |
| Rank30 | ModelDE | Score73.20% | Percentile53.23% | Participants63 | EvidenceC | Evaluated |
BrowseComp Highlights
The leading models and scores on this benchmark.
BrowseComp Score Distribution
A closer view of the leading scores on this benchmark.
The Top AI Models for BrowseComp
The first five results on this benchmark, with official price and output speed added where the model identity can be matched.
What is BrowseComp?
What BrowseComp measures and how its scores work.
BrowseComp is a benchmark of 1,266 questions requiring AI agents to search the internet for concise, verifiable answers.
It measures agents' persistence in information gathering, creativity in web navigation, and ability to find concise, verifiable answers.
Scores are shown in ratio. This benchmark is not independently verified and has an evidence level of B.
- Family
- BrowseComp
- Modality
- text
- Primary category
- reasoning
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- overall
LLM Stats. Benchmark scores retain their original unit. Overall score eligibility is shown separately.
FAQ
Common questions about BrowseComp.
Which model scores highest on BrowseComp?
GPT-6 Astra is currently ranked first with 91.50%.
What are the top three models on BrowseComp?
The current leaders are GPT-6 Astra (91.50%), Kimi K3 (91.20%), and Claude Opus 5 (90.80%).
Which BrowseComp model has the lowest official input price?
Step-3.5-Flash has the lowest matched official input price at $0.10 input / $0.30 output per 1M tokens.
Which models are fastest among BrowseComp results?
The fastest matched records are GPT-5.5 (134.94 tok/s via OpenAI), Claude Opus 4.6 (123.15 tok/s via Anthropic), and o4-mini (115.00 tok/s via OpenAI).
Does the highest score result prove overall model quality?
No. This benchmark measures one defined capability or task. The overall LLMBoard score uses a separate aggregation across eligible benchmark evidence.