Datasets:

Lexmount
/

LexBench-Browser

license: apache-2.0
language:
  - zh
  - en
tags:
  - browser-agent
  - web-automation
  - benchmark
  - evaluation
size_categories:
  - n<1K
task_categories:
  - question-answering
  - text-generation
configs:
  - config_name: LexBench-Browser
    data_files:
      - split: l1
        path: LexBench-Browser/l1.jsonl

LexBench-Browser

A public subset of LexBench-Browser benchmark for evaluating AI agents on web browsing tasks.

Dataset Description

LexBench-Browser is a benchmark designed to evaluate AI Agents on real-world web browsing tasks. This public subset contains 187 tasks that do not require login credentials, making them suitable for open evaluation.

Dataset Statistics

Attribute	Value
Total Tasks	187
Version	2.0
Scenario Tier	L1 (No login required)
Languages	Chinese (115), English (72)

Domain Distribution

Domain	Count
ecommerce	25
finance_gaming	41
general	6
social_lifestyle	28
tools_education	44
video_platform	43

Download

Using Hugging Face Datasets

from datasets import load_dataset

dataset = load_dataset("Lexmount/LexBench-Browser")

# Access a specific split
l1_tasks = dataset["l1"]
print(l1_tasks[0])

For Users in China (Mirror)

If you have difficulty accessing Hugging Face, use the mirror:

import os
os.environ['HF_ENDPOINT'] = 'https://hf-mirror.com'

from datasets import load_dataset
dataset = load_dataset("Lexmount/LexBench-Browser")

Or set environment variable before running:

export HF_ENDPOINT=https://hf-mirror.com
python your_script.py

Manual Download

# Install huggingface-cli if not installed
pip install huggingface_hub

# Download dataset
huggingface-cli download Lexmount/LexBench-Browser --repo-type dataset --local-dir ./data

For China users:

export HF_ENDPOINT=https://hf-mirror.com
huggingface-cli download Lexmount/LexBench-Browser --repo-type dataset --local-dir ./data

Task Format

Each task contains the following fields:

{
  "id": 1,
  "query": "Task description",
  "scenario_tier": "L1",
  "task_type": "T1",
  "reasoning_type": "multi_step",
  "domain": "ecommerce",
  "difficulty": "medium",
  "login_required": false,
  "target_website": "www.example.com",
  "language": "zh",
  "website_region": "zh",
  "reference_answer": {
    "steps": ["Step 1", "Step 2"],
    "key_points": ["Key point 1"],
    "common_mistakes": ["Common mistake 1"],
    "scoring": {
      "total": 100,
      "items": [{"name": "...", "score": 30, "description": "..."}],
      "deductions": [{"reason": "...", "penalty": 20}]
    }
  }
}

Field Descriptions

scenario_tier: L1 - Tasks that do not require login
task_type: T1 (Information retrieval) or T2 (Operation execution)
reasoning_type: single_step, multi_step, cross_platform, or deep_analysis
domain: Business domain (ecommerce, social_lifestyle, video_platform, etc.)
difficulty: easy, medium, or hard
language: Task description language (zh for Chinese, en for English)
website_region: Target website region (zh for Chinese sites, en for international sites)

Related Resources

LexBench-Browser-Private - Private subset requiring login (L2/L3)
LexBench-Online-Mind2Web - Online Mind2Web benchmark

License

Apache 2.0

Citation

If you use this dataset, please cite:

@misc{lexbench-browser-2026,
  title={LexBench-Browser: A Benchmark for Web Browsing AI Agents},
  author={Lexmount},
  year={2026},
  publisher={Hugging Face},
  url={https://huggingface.co/datasets/Lexmount/LexBench-Browser}
}