Your benchmark score is a harness score
GLM-5.3-Flash · Terminal-Bench 2.1 · one model, many harnesses
0
/ 0 trials
0
running ·
0
people contributing
Workshop leaderboard · Ref = Harness Leaderboard
Harness
Pass rate
95% CI
Ref
Trials
$ / trial
Cost vs pass rate
workshop runs (≥ 5 trials)
Harness Leaderboard (ref)
Grid ·
open
running
pass
fail
Newest results