HAL TAU-bench Airline: leaderboard
Princeton HAL cost-aware agent leaderboard for TAU-bench Airline customer-service tool-use tasks.
Metric: Accuracy (%). Source: hal.cs.princeton.edu. Saturation forecast: Around January 2027. 26 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude 3.7 Sonnet (20250219) | 56 | #201 |
| 2 | O4 Mini (High) | 56 | #180 (O4 Mini) |
| 3 | Claude Opus 4.1 (20250805) | 54 | #132 |
| 4 | O3 (Medium) | 54 | #124 (O3) |
| 5 | GPT-5 (Medium) | 48 | #98 (GPT-5) |
| 6 | Claude Opus 4 (20250514) | 44 | #154 |
| 7 | DeepSeek R1 | 36 | #251 |
| 8 | GPT-4.1 (2025-04-14) | 36 | #250 |
| 9 | O4 Mini (Low) | 36 | #180 (O4 Mini) |
| 10 | Gemini 2.0 Flash (001) | 22 | #360 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=hal-tau-bench-airline · How It Works · Data refreshed daily, snapshot 2026-10-20.