Benchmark
tau-bench original retail pass^1
Agentic benchmarks measure how well models carry out tasks that take several steps. Scores are percentages. Higher is better.
- Models measured
- 3
- Scores recorded
- 3
- Sources
- 1
- IQ weight
- No weight published
Score distribution
- 81.4%
- Mean
- 81.4%
- Median
- 0.776%
- Std dev
- 80.5%
- 10th percentile
- 82.4%
- 90th percentile
- 80.5%
- Min
Leaderboard
| # | Model | Score |
|---|---|---|
| 1 | Claude Opus 4.1submission-claude-opus-4-1_anthropic_2025-01-15· Sierra tau-bench Leaderboard data docs | 82.4% |
| 2 | Claude Opus 4submission-claude-opus-4_anthropic_2025-05-22· Sierra tau-bench Leaderboard data docs | 81.4% |
| 3 | Claude Sonnet 4submission-claude-sonnet-4_anthropic_2025-05-22· Sierra tau-bench Leaderboard data docs | 80.5% |
Data credits: Sierra Research tau2-bench leaderboard submissions (github.com/sierra-research/tau2-bench), MIT (licence)