Skip to content

Benchmark

Terminal-Bench · Hard 48

Agentic benchmarks measure how well models carry out tasks that take several steps. Scores are percentages. Higher is better.

Models measured
1
Scores recorded
2
Sources
1
IQ weight
No weight published

Score distribution

25.8%
Mean
25.8%
Median
0.01%
Std dev
25.8%
10th percentile
25.8%
90th percentile
25.8%
Min
25.78%25.8%
The band covers the middle 80% of scores. The line marks the median.

Leaderboard

#ModelScore
1Nemotron 3 Super 120B A12Bofficial blog Figure 3 · rounded· Reported by NVIDIA Developer· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.25.8%
2Nemotron 3 Super 120B A12BNeMo Evaluator / mostly NeMo Skills · dedicated open-source container · 48 tasks· Reported by NVIDIA· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.25.8%

Data credits: BenchBoard by SYuan03 (github.com/SYuan03/benchboard). MIT covers its code and data organization; no results licence is inferred. Numeric relays retain the original stating-source links.