Skip to content

docs: add related ClawBench web-agent benchmark #14

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
reacher-z wants to merge 2 commits into
base: main
Choose a base branch
from
Open

docs: add related ClawBench web-agent benchmark #14

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
19bf591
docs: add related ClawBench benchmark
reacher-z Jul 28, 2026
7a8b5a0
fix: use canonical ClawBench paper link
reacher-z Jul 28, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 4 additions & 0 deletions README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -54,6 +54,10 @@ Current research primarily focuses on "deep" tasks. DeepSearch tackles the "I ca

In sharp contrast, WideSearch tackles the "I could do it, but the sheer volume is overwhelming" problem. It requires agents to systematically find and organize large-scale information into a structured output, shifting the primary challenge from deep search to achieving exhaustiveness and fidelity at scale.

### Related Web-Agent Benchmark

[ClawBench](https://github.com/TIGER-AI-Lab/ClawBench) is a complementary benchmark for evaluating agents on long-horizon, real-world browser tasks across live websites. It emphasizes end-to-end execution, browser actions, network traces, screenshots, and rubric-based evaluation, providing a useful contrast to WideSearch's broad information-seeking focus. See the [project homepage](https://claw-bench.com/) and [paper](https://arxiv.org/abs/2604.08523) for details.

## Experiments
We test both single-agent and multi-agent modes, and manually conducted end-to-end testing of the commercial AI system on the web interface. In addition, we randomly select 20 questions and invited human annotators to perform tests. The experiment results are as follows:
![experiments](figs/image.png)
Expand Down