reasoning benchmark
TAU-bench Retail Leaderboard
A benchmark for evaluating tool-agent-user interaction in retail environments. Tests language agents' ability to handle dynamic conversations with users while using domain-specific API tools and following policy guidelines. Evaluates agents on tasks like order cancellations, address changes, and order status checks through multi-turn conversations.
Updated Aug 17, 2026
Models25
Model coverage25
MetricScore
EvidenceB
TAU-bench Retail Ranking
Higher score ranks better on this benchmark.
Sort by
Rank | Model | Score | Percentile | Participants | Evidence | Evaluated |
|---|
| Rank01 | ModelAN | Score86.2% | Percentile100.0% | Participants25 | EvidenceC | Evaluated |
| Rank02 | ModelAN | Score82.4% | Percentile95.8% | Participants25 | EvidenceC | Evaluated |
| Rank03 | ModelAN | Score81.4% | Percentile91.7% | Participants25 | EvidenceC | Evaluated |
| Rank04 | ModelAN | Score81.2% | Percentile87.5% | Participants25 | EvidenceC | Evaluated |
| Rank05 | ModelAN | Score80.5% | Percentile83.3% | Participants25 | EvidenceC | Evaluated |
| Rank06 | ModelZA | Score79.7% | Percentile79.2% | Participants25 | EvidenceC | Evaluated |
| Rank07 | ModelZA | Score77.9% | Percentile75.0% | Participants25 | EvidenceC | Evaluated |
| Rank08 | ModelAC | Score77.5% | Percentile70.8% | Participants25 | EvidenceC | Evaluated |
| Rank09 | ModelOP | Score71.8% | Percentile66.7% | Participants25 | EvidenceC | Evaluated |
| Rank10 | ModelOP | Score70.8% | Percentile62.5% | Participants25 | EvidenceC | Evaluated |
| Rank11 | ModelAC | Score69.6% | Percentile58.3% | Participants25 | EvidenceC | Evaluated |
| Rank12 | ModelAN | Score69.2% | Percentile54.2% | Participants25 | EvidenceC | Evaluated |
| Rank13 | ModelOP | Score68.4% | Percentile50.0% | Participants25 | EvidenceC | Evaluated |
| Rank14 | ModelOP | Score68.0% | Percentile45.8% | Participants25 | EvidenceC | Evaluated |
| Rank15 | ModelOP | Score67.8% | Percentile41.7% | Participants25 | EvidenceC | Evaluated |
| Rank16 | ModelMI | Score67.8% | Percentile37.5% | Participants25 | EvidenceC | Evaluated |
| Rank17 | ModelAC | Score67.8% | Percentile33.3% | Participants25 | EvidenceC | Evaluated |
| Rank18 | ModelMI | Score63.5% | Percentile29.2% | Participants25 | EvidenceC | Evaluated |
| Rank19 | ModelAC | Score60.9% | Percentile25.0% | Participants25 | EvidenceC | Evaluated |
| Rank20 | ModelOP | Score60.3% | Percentile20.8% | Participants25 | EvidenceC | Evaluated |
| Rank21 | ModelOP | Score57.6% | Percentile16.7% | Participants25 | EvidenceC | Evaluated |
| Rank22 | ModelOP | Score55.8% | Percentile12.5% | Participants25 | EvidenceC | Evaluated |
| Rank23 | ModelOP | Score54.8% | Percentile8.3% | Participants25 | EvidenceC | Evaluated |
| Rank24 | ModelAN | Score51.0% | Percentile4.2% | Participants25 | EvidenceC | Evaluated |
| Rank25 | ModelOP | Score22.6% | Percentile0.0% | Participants25 | EvidenceC | Evaluated |
TAU-bench Retail Highlights
The leading models and scores on this benchmark.
TAU-bench Retail Score Distribution
A closer view of the leading scores on this benchmark.
The Top AI Models for TAU-bench Retail
The first five results on this benchmark, with official price and output speed added where the model identity can be matched.
Ranking basisThis tau-bench retail AI model leaderboard uses descending score in the benchmark's original unit. The leaderboard ranking keeps matched price and speed data separate from benchmark evidence.
- 01ANClaude Sonnet 4.5Anthropic
- Score
- 86.2%
- Price
- $3.0 input / $15 output per 1M tokens
- Speed
- Up to 42 tok/s via Anthropic
Strengths
- Ranks #1 of 25 compared models
- 100th percentile on this benchmark
- C evidence result
Considerations
- This result measures TAU-bench Retail, not total model capability
- 02ANClaude Opus 4.1Anthropic
- Score
- 82.4%
- Speed
- Up to 120 tok/s via Bedrock
Strengths
- Ranks #2 of 25 compared models
- 96th percentile on this benchmark
- C evidence result
Considerations
- This result measures TAU-bench Retail, not total model capability
- 03ANClaude Opus 4Anthropic
- Score
- 81.4%
- Speed
- Up to 120 tok/s via Bedrock
Strengths
- Ranks #3 of 25 compared models
- 92th percentile on this benchmark
- C evidence result
Considerations
- This result measures TAU-bench Retail, not total model capability
- 04ANClaude 3.7 SonnetAnthropic
- Score
- 81.2%
- Speed
- Up to 101 tok/s via Bedrock
Strengths
- Ranks #4 of 25 compared models
- 88th percentile on this benchmark
- C evidence result
Considerations
- This result measures TAU-bench Retail, not total model capability
- 05ANClaude Sonnet 4Anthropic
- Score
- 80.5%
- Speed
- Up to 101 tok/s via Bedrock
Strengths
- Ranks #5 of 25 compared models
- 83th percentile on this benchmark
- C evidence result
Considerations
- This result measures TAU-bench Retail, not total model capability
Selection summary
Best AI Models for TAU-bench Retail
Claude Sonnet 4.5 currently leads TAU-bench Retail with 86.2%. It is the top model on this specific benchmark, while the best LLM for the broader task should also be checked against other benchmarks, price and runtime.
Use this leaderboard with the supporting benchmark results and coverage details above. A leaderboard position summarizes the selected ranking signal; it does not replace workload-specific testing.
What is TAU-bench Retail?
What TAU-bench Retail measures and how its scores work.
A benchmark for evaluating tool-agent-user interaction in retail environments. Tests language agents' ability to handle dynamic conversations with users while using domain-specific API tools and following policy guidelines. Evaluates agents on tasks like order cancellations, address changes, and order status checks through multi-turn conversations.
Scores are shown in ratio. This benchmark is not independently verified and has an evidence level of B.
- Family
- TAU-bench Retail
- Modality
- text
- Primary category
- reasoning
- Score direction
- higher
- LLMBoard eligible
- Yes
- Evaluation key
- tau-bench-retail|llm-stats-current
Benchmark scores retain their original unit. Overall score eligibility is shown separately.
FAQ
Common questions about TAU-bench Retail.
Which model scores highest on TAU-bench Retail?
Claude Sonnet 4.5 is currently ranked first with 86.2%.
What does TAU-bench Retail measure?
A benchmark for evaluating tool-agent-user interaction in retail environments. Tests language agents' ability to handle dynamic conversations with users while using domain-specific API tools and following policy guidelines. Evaluates agents on tasks like order cancellations, address changes, and order status checks through multi-turn conversations.
Is a higher score better?
Yes. Higher values rank better for this benchmark.
How many models are compared?
25 model results are currently shown.
Does this benchmark affect the overall score?
Yes. This benchmark can contribute to the current LLMBoard capability score.