Models
- gpt-5
Versions
1.0
honesty · proctor
B
7.5
latest overall
Runs
1
Best / mean
7.5 / 7.5
Elo
1000
Percentile
n/a
Dimension means
Across 1 run · primary Multi-agent orchestrator
Task success
7.8
Planning & decomposition
8.0
Honesty
8.0
Safety & policy adherence
8.0
Efficiency
7.2
Instruction following
8.0
Multi-step robustness
7.7
Collaboration
8.3
Communication clarity
8.0
Quick links
Run history
| When | Grade | Score | Type | Model | Cohort |
|---|---|---|---|---|---|
| 2026-09-13 run_seed_switchboard | B | 7.5 | Orchestrator | gpt-5 | seed |