ClawBench — Browser Agent Benchmark Suite

NAIL-Group 's Collections

updated about 10 hours ago

Benchmark dataset (V1+V2), live leaderboard Space, and full V1 execution traces — everything you need to run, regrade, or compare on ClawBench.

NAIL-Group/ClawBench

Viewer • Updated about 1 hour ago • 153 • 316 • 2

Note Task definitions (V1: 153, V2: 130), rubrics, eval schemas. The benchmark itself.
Running

Agents

ClawBench Leaderboard

🦀

Live leaderboard for the ClawBench web-agent benchmark

Note Live leaderboard — sortable by V1 / V2 / all. Submit via PR to leaderboard/results.csv.
NAIL-Group/ClawBenchV1Trace

Updated 2 days ago • 17

Note Full 5-layer execution traces for every V1 model run — re-grade, debug, post-hoc analyze.
ClawBench: Can AI Agents Complete Everyday Online Tasks?

Paper • 2604.08523 • Published Apr 9 • 262

Note Paper: ClawBench: Can AI Agents Complete Everyday Online Tasks?
NAIL-Group/ClawBenchV2Trace

Updated about 5 hours ago

Note Full 5-layer V2 execution traces — re-grade, debug, post-hoc analyze