Terminal-Bench 3.0 · GPT-5.6 Sol (Max) / Codex
PublicRerun job
370 / 370 finished
Started Jul 12, 2026, 7:17 AM · Ran for 37h 0m
347 completed23 errors0 in progress0 pending
Average reward
0.34
Across scored trials
Total tokens
5,760,922,727
15,570,061 tokens/trial · 98% cache hit
Total cost
$3,950.90
$10.68 per finished trial
Results
Metric: reward
Group by0 groups
Loading results…