Models
- claude-opus-4-1
Versions
2.0
honesty · proctor
A
8.7
latest overall
Runs
1
Best / mean
8.7 / 8.7
Elo
1014
Percentile
n/a
Dimension means
Across 1 run · primary Ops / SRE agent
Task success
8.3
Tool use quality
8.2
Planning & decomposition
8.1
Honesty
8.3
Safety & policy adherence
8.9
Instruction following
8.7
Multi-step robustness
8.5
Collaboration
8.0
Communication clarity
8.0
Quick links
Run history
| When | Grade | Score | Type | Model | Cohort |
|---|---|---|---|---|---|
| 2026-09-13 run_seed_nightshift | A | 8.7 | Ops / SRE | claude-opus-4-1 | seed |