0

BrowseComp

Frontier

1,266 hard fact-finding questions on the open web requiring persistent browsing and reasoning over scattered, obscure sources.

Open
Publisher
OpenAI
Capabilities
Browser UseRetrievalPlanning
Domain
agentic
Format
Custom
Size
1266 tasks
License
MIT
Published
Apr 2025
Updates
Monthly
Notable for
The canonical benchmark for "deep research" / browsing agents — where GPT-4o scored near 0% at launch but GPT-5.5 Pro now exceeds 90%.
Canonical
openai.com/index/browsecomp
Official leaderboard
openai.com/index/browsecomp
Also on
github.com/openai/simple-evals

Cite

Notes

Only stored in your browser.

Attribution

Leaderboard scores
prime-hub
Attribution policy →

Top score 20.0% by GPT-5 Mini - 5 models reporting (5 frontier)

Score history

5
0%25%50%75%100%Apr 25Jun 25Aug 25Oct 25Dec 25o3GPT-5 Mini

Top models

5
BrowseCompBar chart with 5 bars. Highest value: GPT-5 Mini at 20.
5 models

Where it's ranked

1

Official leaderboard

openai.com

Single benchmark
monthly

Related tools

7
View all

Implementations, trainers, datasets and scaffolds linked to this eval.

BB Browsecomp RL Env (Prime Intellect)

Prime Intellect

BrowserEnv[DOM] Environment on Browsecomp

ImplementationRL Env

Browsecomp RL Env (Prime Intellect)

Prime Intellect

BrowseComp evaluation environment

ImplementationRL EnvTool UseWeb SearchWeb

DeepDive (Serper-powered web QA)

Prime Intellect

Multi-turn open-web research QA environment where the model uses a Serper search tool to answer hard, hop-heavy questions of the BrowseComp / SimpleQA style.

Trains towardRL EnvPlanningTool CallingRetrieval

Browsecomp Openai RL Env (Community)

Tool-use environment for the model to browse the web and locate hard-to-find information; scored using an LLM-as-judge rubric

Trains towardRL EnvTool UseWeb SearchWeb

Browsecomp Openai RL Env (Prime Intellect)

Prime Intellect

Tool-use environment for the model to browse the web and locate hard-to-find information; scored using an LLM-as-judge rubric

Trains towardRL EnvTool UseWeb SearchWeb

DDBC RL Env (Prime Intellect)

Prime Intellect

browsecomp with deepdive tools

Trains towardRL EnvQASearch

DDBC RLM RL Env (Prime Intellect)

Prime Intellect

BrowseComp with DeepDive tools using RLM

Trains towardRL EnvQASearch

Papers

2

BrowseComp: A Simple Yet Challenging Benchmark for Browsing Agents

preprint · 2025

OpenAI benchmark of 1,266 web-research questions that require persistent, creative browsing to find a single short verifiable answer.

introduces

BrowseComp: A Simple Yet Challenging Benchmark for Browsing Agents

preprint · 2025

OpenAI benchmark of 1,266 web-research questions that require persistent, creative browsing to find a single short verifiable answer.

Contributors

1
JJason Wei

FAQ

What is BrowseComp?
1,266 hard fact-finding questions on the open web requiring persistent browsing and reasoning over scattered, obscure sources.
What capabilities does BrowseComp test?
BrowseComp evaluates browser use, retrieval, planning.
What is the current top score on BrowseComp?
The top reported score is 20.0% by GPT-5 Mini, across 5 models reporting (5 from frontier labs).
How can a model improve its BrowseComp score?
Tools linked to BrowseComp on Sophon include BB Browsecomp RL Env (Prime Intellect), Browsecomp RL Env (Prime Intellect), DeepDive (Serper-powered web QA), Browsecomp Openai RL Env (Community) - RL environments, datasets, and scaffolds that target this eval.
What license is BrowseComp under?
BrowseComp is available under MIT.