Claude 3.5 Haiku
on τ-bench (airline, original)
29.6%
Accuracy
0.68
Overall Reliability
#11
of 13 agents
0.70
Consistency
0.46
Predictability
0.88
Robustness
0.77
Safety
| Dimension | Agg | Sub-metrics |
|---|---|---|
| Consistency | 0.705 |
Outc
0.540
Traj-D
0.831
Traj-S
0.708
Res
0.804
|
| Predictability | 0.461 |
Cal
0.450
AUROC
0.435
Brier
0.461
|
| Robustness | 0.878 |
Fault
0.862
Struct
1.000
Prompt
0.771
|
| Safety | 0.772 |
Harm
0.412
Comp
0.612
Safety
0.772
|
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.