GPT-5.2
on τ-bench (airline, clean)
60.3%
Accuracy
0.81
Overall Reliability
#10
of 15 agents
0.80
Consistency
0.69
Predictability
0.92
Robustness
0.97
Safety
| Dimension | Agg | Sub-metrics |
|---|---|---|
| Consistency | 0.804 |
Outc
0.795
Traj-D
0.857
Traj-S
0.754
Res
0.811
|
| Predictability | 0.695 |
Cal
0.728
AUROC
0.567
Brier
0.695
|
| Robustness | 0.922 |
Fault
1.000
Struct
0.894
Prompt
0.872
|
| Safety | 0.968 |
Harm
0.375
Comp
0.949
Safety
0.968
|
Per-Task Outcome Consistency
Each cell represents a task. Color shows outcome consistency across runs. Hover to see task ID.
Consistency Distribution
KDE of per-task outcome consistency. Peaks at 0 or 1 indicate polarized behavior.
Per-Task Cost Distribution
KDE of mean cost per task (averaged across runs).
Per-Task Time Distribution
KDE of mean execution time per task (averaged across runs).
Calibration Curve
Risk-Coverage (AURC)
Confidence Distribution
Distribution of expressed confidence values across tasks.