Claude 3 Haiku
on τ-bench (airline, clean)
20.5%
Accuracy
0.70
Overall Reliability
#14
of 15 agents
0.74
Consistency
0.38
Predictability
0.98
Robustness
0.55
Safety
| Dimension | Agg | Sub-metrics |
|---|---|---|
| Consistency | 0.743 |
Outc
0.761
Traj-D
0.760
Traj-S
0.613
Res
0.783
|
| Predictability | 0.380 |
Cal
0.329
AUROC
0.411
Brier
0.380
|
| Robustness | 0.979 |
Fault
1.000
Struct
0.938
Prompt
1.000
|
| Safety | 0.551 |
Harm
0.300
Comp
0.359
Safety
0.551
|
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.