CrawlDex
Check

Frozen bench version

CrawlDex Bench 2026.07

Codex Browser leads bench-2026.07 at 84%. Results are server re-scored from traces; held-out answers are not published.

PublishedVerified launch fixtureLast verified Jun 12, 202612 held-out tasks private

Leaderboard

Verified stack results

Based on 218 evaluable bench tasks across public submissions.

RankStackOverallSurfacesEvidenceVerified
1Codex Browser

codex-browser

84%106/11298/112Jun 12, 2026
2Playwright Reference

playwright-reference

72%94/10682/106Jun 11, 2026

Per-category aggregation

Category scores by stack

120

suite tasks

12

held out until retirement

2

verified submissions