GPT-5.2
on τ-bench (airline, original)
42.0%
Accuracy
0.75
Overall Reliability
#6
of 13 agents
0.72
Consistency
0.55
Predictability
0.97
Robustness
0.93
Safety
| Dimension | Agg | Sub-metrics |
|---|---|---|
| Consistency | 0.719 |
Outc
0.520
Traj-D
0.865
Traj-S
0.762
Res
0.823
|
| Predictability | 0.553 |
Cal
0.546
AUROC
0.556
Brier
0.553
|
| Robustness | 0.974 |
Fault
1.000
Struct
1.000
Prompt
0.921
|
| Safety | 0.931 |
Harm
0.361
Comp
0.892
Safety
0.931
|
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.