Terminal-bench
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
Terminal-Bench 2.0
26.2Accuracy
7
Terminal Bench 2.1
67.2Score
7
Terminal-Bench 2.0
75.1Public Score
7
Terminal-Bench
53.8Score
7
Terminal-Bench Hard 2 (30 tasks)
56.7Pass@1
7
Terminal-Bench Med. 2 (55 tasks)
88.2Pass@1
7
Terminal-Bench Easy 4 tasks 2
100Pass@1
7
Terminal-Bench All 2
77Pass@1
7
Terminal Bench
23Pass@1
7
Terminal-Bench 2.0
57.3Terminus2 JSON Score
7
Terminal Bench Hard
36Accuracy
6
Terminal-Bench 2.0 (full)
58.9Overall avg@5 Accuracy
6
Terminal-Bench out-of-distribution 2.0 (test)
39.4Avg@5
6
Terminal Bench 2.0
54.2Accuracy
6
Terminal-Bench 1.0 (test)
34.9Avg Pass Rate
6
Terminal Bench hard
25.5Accuracy
5
Terminal Bench 2.0
62.4Score
5
Terminal Bench 2.1
82.1Score
5
Terminal-Bench 2 (held-out)
76Pass Rate
5
Terminal-Bench V2
86.5Accuracy
4
Terminal-Bench agentic 2.0
74.2Quality
4
Terminal-Bench 2.0 (test)
77.36Pass Rate
4
Terminal-Bench 2.0 (dev)
75Pass Rate
4
Terminal Bench 77 tasks (held-out)
100Hinted ASR
4
Terminal Bench (held-out)
41.4Trajectory-level ASR
4