Skip to content

Benchmark

Terminal-Bench · Claude 4 launch snapshot

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

Models measured
2
Scores recorded
6
Sources
1
IQ weight
No weight published

Score distribution

40.5%
Mean
41.3%
Median
5.4%
Std dev
35.5%
10th percentile
50%
90th percentile
33.5%
Min
33.5%50%
The band covers the middle 80% of scores. The line marks the median.

Leaderboard

#ModelScore
1Claude Opus 4Claude Code agent · parallel test-time compute· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.50%
2Claude Opus 4pass@1 · Claude Code agent · no extended thinking· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.43.2%
3Claude Sonnet 4Claude Code agent · parallel test-time compute· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.41.3%
4Claude Opus 4pass@1 · same agent as non-Claude comparison models · no extended thinking· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.39.2%
5Claude Sonnet 4pass@1 · Claude Code agent · no extended thinking· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.35.5%
6Claude Sonnet 4pass@1 · same agent as non-Claude comparison models · no extended thinking· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.33.5%

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.