Terminal-Bench 4.0.0 · GPT-5.6 Sol / Codex
PublicRerun job
330 / 330 finished
Started Aug 26, 2026, 6:18 PM · Ran for 28h 13m
324 completed4 timeouts2 errors0 in progress0 pending
Average reward
0.37
Across scored trials
Total tokens
4,539,672,216
13,756,582 tokens/trial · 98% cache hit
Inference cost
$2,605.19
$7.89 per finished trial
Sandbox cost
Pending
Costs appear after usage is processed
Results
Metric: reward
Group by0 groups
Loading results…