GPT-5.5
on τ-bench (airline, clean)
79.5%
Accuracy
0.89
Overall Reliability
#2
of 15 agents
0.84
Consistency
0.84
Predictability
0.97
Robustness
0.96
Safety
| Dimension | Agg | Sub-metrics |
|---|---|---|
| Consistency | 0.839 |
Outc
0.829
Traj-D
0.883
Traj-S
0.815
Res
0.840
|
| Predictability | 0.844 |
Cal
0.890
AUROC
0.720
Brier
0.844
|
| Robustness | 0.973 |
Fault
1.000
Struct
0.919
Prompt
1.000
|
| Safety | 0.962 |
Harm
0.400
Comp
0.936
Safety
0.962
|
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.