Gemini 2.0 Flash
on τ-bench (airline, clean)
44.6%
Accuracy
0.70
Overall Reliability
#11
of 13 agents
0.65
Consistency
0.48
Predictability
0.98
Robustness
0.87
Safety
| Dimension | Agg | Sub-metrics |
|---|---|---|
| Consistency | 0.651 |
Outc
0.346
Traj-D
0.885
Traj-S
0.729
Res
0.801
|
| Predictability | 0.475 |
Cal
0.462
AUROC
0.565
Brier
0.475
|
| Robustness | 0.984 |
Fault
0.976
Struct
1.000
Prompt
0.977
|
| Safety | 0.867 |
Harm
0.405
Comp
0.777
Safety
0.867
|
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.