Third-party content
The Apache-2.0 license covers project-owned code, documentation and task requirements. Initial definitions and input-checking concepts originate in Browser Eval; its LexMount copyright notice is retained in NOTICE.
Reference texts included for evaluation
definitions/materials/primary-content-20261103/ includes the reference texts that scoring needs for all five suites (webfetch, dynamic-content, popular-sites, access-restrictions and articles): main-content references, reference body regions and article endings, kept in their original language and wording. They are included only so the benchmark can be scored from a clean download. The Apache-2.0 license does not cover them. Each text keeps the terms of its source website, and inclusion does not grant any further right to reuse it.
Page captures, screenshots, document and media files, source observation files and measured provider outputs are not included.
If you own rights in an included text and want it removed or changed, open a discussion in this dataset repository's Community tab. Give the task ID and source URL. We will remove or replace the text and record the change in a new revision. Earlier revisions stay in the repository history.
WebMainBench, WCXB and WebArena are separate projects and are not redistributed as LexBench Web Fetch suites. Cite their owners and follow their terms when using them through an execution harness.
Scoring dependencies
Exact versions and resolved hashes are in pyproject.toml and uv.lock.
| Library | Use | License |
|---|---|---|
| Beautiful Soup | HTML parsing | MIT |
| Soup Sieve | CSS selectors | MIT |
| APTED | Optional tree-edit scoring | MIT |
| RapidFuzz | Optional text matching | MIT |
These libraries retain their own licenses. Installing the tree extra enables the optional scoring dependencies.