Robustness on τ-bench (airline, clean)
How well does the agent maintain accuracy when inputs are perturbed (faults, structural changes, prompt rewording)? See full definition →
Sub-metric Comparison
Baseline vs Perturbed Accuracy
Agent Leaderboard — Robustness
| # | Agent | Acc | Reliability | Robustness Agg | Fault | Struct | Prompt |
|---|---|---|---|---|---|---|---|
| 1 |  Claude 3 Haiku |
20.5% | 0.70 | 0.98 | 1.00 | 0.94 | 1.00 |
| 2 |  O1 |
66.2% | 0.81 | 0.98 | 1.00 | 1.00 | 0.93 |
| 3 |  Gemini 3.5 Flash |
80.8% | 0.86 | 0.97 | 1.00 | 1.00 | 0.92 |
| 4 | GPT-5.5 |
79.5% | 0.89 | 0.97 | 1.00 | 0.92 | 1.00 |
| 5 | Claude Opus 4.5 |
80.8% | 0.88 | 0.97 | 1.00 | 0.98 | 0.92 |
| 6 | Claude Sonnet 4 |
78.2% | 0.86 | 0.96 | 0.99 | 0.98 | 0.92 |
| 7 | GPT-5.2 (medium) |
67.9% | 0.82 | 0.95 | 1.00 | 0.85 | 1.00 |
| 8 | Gemini 3.1 Pro |
82.1% | 0.86 | 0.94 | 0.98 | 1.00 | 0.84 |
| 9 | Gemini 2.5 Pro |
71.8% | 0.81 | 0.93 | 1.00 | 1.00 | 0.79 |
| 10 | Claude 3.5 Haiku |
29.5% | 0.71 | 0.93 | 1.00 | 0.78 | 1.00 |
| 11 | GPT-4o Mini |
29.5% | 0.67 | 0.93 | 1.00 | 0.91 | 0.87 |
| 12 | GPT-5.2 |
60.3% | 0.81 | 0.92 | 1.00 | 0.89 | 0.87 |
| 13 | Claude Opus 4.7 |
84.6% | 0.89 | 0.91 | 0.91 | 1.00 | 0.83 |
| 14 | Gemini 2.5 Flash |
59.0% | 0.72 | 0.82 | 0.78 | 0.85 | 0.83 |
| 15 | GPT-4 Turbo |
57.7% | 0.72 | 0.81 | 0.83 | 0.73 | 0.87 |