Benchmark
Tau2 Airline Benchmark
What is Tau2 Airline?
TAU2 airline domain benchmark for evaluating conversational agents in dual-control environments where both AI agents and users interact with tools in airline…
- Released
- 2025-06-09
- Modality
- text
- Evaluates
- reasoning, communication, tool calling
- Openness
- restricted
- Importer
- LLM Stats
- Reported scores
- 23
Source provenance
- Original evidence https://api.zeroeval.com/leaderboard/benchmarks/tau2-airline?top_n=500
Tau2 Airline paper and code
Tau2 Airline results and reported scores
LLM Stats
| Model | Organization | Reported value | Reported | Evidence |
|---|---|---|---|---|
| Claude Haiku 4.5 | Anthropic | 0.636 | 2025-10-15 | evidence |
| GPT-4o | OpenAI | 0.455 | 2024-08-06 | evidence |
| GPT-5 | OpenAI | 0.626 | 2025-08-07 | evidence |
| GPT-5.1 | OpenAI | 0.67 | 2025-11-13 | evidence |
| GPT-5.1 Instant | OpenAI | 0.67 | 2025-11-12 | evidence |
| GPT-5.1 Thinking | OpenAI | 0.67 | 2025-11-12 | evidence |
| Kimi K2 Instruct | Moonshot AI | 0.565 | 2025-07-11 | evidence |
| Kimi K2-Instruct-0905 | Moonshot AI | 0.565 | 2025-09-05 | evidence |
| LongCat-Flash-Chat | Meituan | 0.58 | 2025-08-29 | evidence |
| LongCat-Flash-Lite | Meituan | 0.58 | 2026-02-05 | evidence |
| LongCat-Flash-Thinking | Meituan | 0.675 | 2025-09-22 | evidence |
| LongCat-Flash-Thinking-2601 | Meituan | 0.765 | 2026-01-14 | evidence |
| Mercury 2 | Inception | 0.53 | 2026-02-24 | evidence |
| Nemotron 3 Nano (30B A3B) | NVIDIA | 0.48 | 2025-12-15 | evidence |
| Nemotron 3 Super (120B A12B) | NVIDIA | 0.5625 | 2026-03-11 | evidence |
| Nova 2 Lite | Amazon | 0.648 | 2025-12-02 | evidence |
| Nova 2 Omni | Amazon | 0.688 | 2025-12-02 | evidence |
| Nova 2 Pro | Amazon | 0.652 | 2025-12-02 | evidence |
| o3 | OpenAI | 0.648 | 2025-04-16 | evidence |
| Qwen3-235B-A22B-Instruct-2507 | Qwen | 0.44 | 2025-07-22 | evidence |
| Qwen3-235B-A22B-Thinking-2507 | Qwen | 0.58 | 2025-07-25 | evidence |
| Qwen3-Next-80B-A3B-Instruct | Qwen | 0.455 | 2025-09-10 | evidence |
| Qwen3-Next-80B-A3B-Thinking | Qwen | 0.605 | 2025-09-10 | evidence |
Scores are partitioned by the source that reported them and are never merged into a single cross-source ranking, because the sources measure different things and say so.