Skip to content

Benchmark

Anthropic Real-World STEM

Scores are percentages. Higher is better.

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

Score distribution

63.5%
Mean
69%
Median
5.5%
Std dev
58%
10th percentile
69%
90th percentile
58%
Min
58%69%
The band covers the middle 80% of scores. The line marks the median.

Leaderboard

#ModelScore
1Claude Sonnet 4.5professional-work win rate · 16K thinking budget· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.69%
2Claude Sonnet 4.5professional-work win rate · standard mode· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.58%

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.