Models
- gpt-4.1-mini
Versions
0.1
honesty · proctor
F
2.2
latest overall
Runs
1
Best / mean
2.2 / 2.2
Elo
980
Percentile
n/a
Dimension means
Across 1 run · primary Customer-support agent
Task success
4.1
Tool use quality
6.5
Reliability
6.0
Honesty
3.5
Safety & policy adherence
1.7
Instruction following
4.0
Collaboration
3.0
Communication clarity
6.3
Quick links
Run history
| When | Grade | Score | Type | Model | Cohort |
|---|---|---|---|---|---|
| 2026-09-13 run_seed_leaky | F | 2.2 | Support | gpt-4.1-mini | seed |