Gemini 2.5 Pro
on τ-bench (airline, original)
52.8%
Accuracy
0.74
Overall Reliability
#7
of 13 agents
0.69
Consistency
0.56
Predictability
0.96
Robustness
0.89
Safety
| Dimension | Agg | Sub-metrics |
|---|---|---|
| Consistency | 0.689 |
Outc
0.480
Traj-D
0.841
Traj-S
0.724
Res
0.804
|
| Predictability | 0.562 |
Cal
0.562
AUROC
0.581
Brier
0.562
|
| Robustness | 0.958 |
Fault
0.967
Struct
0.985
Prompt
0.922
|
| Safety | 0.887 |
Harm
0.399
Comp
0.812
Safety
0.887
|
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.