Skip to content

Benchmark

τ-bench Retail

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

Models measured
4
Scores recorded
8
Sources
1
IQ weight
No weight published

Score distribution

59.8%
Mean
62%
Median
15.3%
Std dev
47.3%
10th percentile
80.5%
90th percentile
35%
Min
35%81.4%
The band covers the middle 80% of scores. The line marks the median.

Leaderboard

#ModelScore
1Claude Opus 4extended thinking + tool use · prompt addendum · max 100 model-completion steps· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.81.4%
2Claude Sonnet 4extended thinking + tool use · prompt addendum · max 100 model-completion steps· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.80.5%
3GPT OSS 120Bofficial model card Table 3 · reasoning high · developer functions · pass@1· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.67.8%
4GPT OSS 120Bofficial model card Table 3 · reasoning medium · developer functions · pass@1· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.62%
5GPT OSS 20bofficial model card Table 3 · reasoning high · developer functions · pass@1· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.54.8%
6GPT OSS 120Bofficial model card Table 3 · reasoning low · developer functions · pass@1· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.49.4%
7GPT OSS 20bofficial model card Table 3 · reasoning medium · developer functions · pass@1· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.47.3%
8GPT OSS 20bofficial model card Table 3 · reasoning low · developer functions · pass@1· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.35%

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.