Compare

Twelve baseline models, one behavior map

This page is not a ranking table. It is a clean side-by-side view of how each model distributes its behavior across the five dimensions, with consistent spacing, hover states, and card structure.

Use it to compare behavioral shape, stability, and collapse tendency before you decide which baseline feels closest to your own archive.
Model

Claude Opus 4.6

Stability 89%
Leading Dimension
Fairness
Care89.2
Fairness92.4
Loyalty49.4
Authority88.8
Sanctity84.6
No obvious collapse dimension in the current benchmark
Model

Claude Sonnet 4.6

Stability 83%
Leading Dimension
Fairness
Care75.6
Fairness92.2
Loyalty46.4
Authority88
Sanctity88.2
Most brittle under pressure: Loyalty
Model

Deepseek R1

Stability 78%
Leading Dimension
Authority
Care88.2
Fairness81.6
Loyalty62.8
Authority89.4
Sanctity70
Most brittle under pressure: Fairness
Model

Deepseek V3.2

Stability 79%
Leading Dimension
Authority
Care87.6
Fairness88.2
Loyalty62
Authority89.6
Sanctity76.8
Most brittle under pressure: Fairness
Model

Gemini 2.5 Pro

Stability 85%
Leading Dimension
Care
Care71
Fairness68
Loyalty50
Authority65.4
Sanctity64
No obvious collapse dimension in the current benchmark
Model

Gemini 3 Flash

Stability 77%
Leading Dimension
Care
Care87.6
Fairness86.2
Loyalty47.6
Authority74.8
Sanctity85
Most brittle under pressure: Sanctity
Model

GPT-5.4

Stability 84%
Leading Dimension
Care
Care93
Fairness90
Loyalty55.8
Authority81.6
Sanctity91.6
Most brittle under pressure: Loyalty
Model

Kimi K2.5

Stability 74%
Leading Dimension
Sanctity
Care72
Fairness69.4
Loyalty42.4
Authority50
Sanctity82.4
Most brittle under pressure: Authority
Model

Llama 4 Maverick

Stability 73%
Leading Dimension
Fairness
Care81.6
Fairness83.4
Loyalty49.6
Authority76
Sanctity70.6
Most brittle under pressure: Authority
Model

MiniMax M2.5

Stability 75%
Leading Dimension
Fairness
Care84.4
Fairness85.2
Loyalty55.2
Authority77.8
Sanctity83.8
Most brittle under pressure: Sanctity
Model

MiniMax M2.7

Stability 75%
Leading Dimension
Fairness
Care77.6
Fairness89.6
Loyalty65.4
Authority82.6
Sanctity77.2
Most brittle under pressure: Sanctity
Model

Qwen3 235B

Stability 80%
Leading Dimension
Care
Care83.6
Fairness64
Loyalty54.4
Authority67.2
Sanctity63.6
Most brittle under pressure: Fairness
Data Note

Baseline gallery only

This page compares the twelve shared baselines. Your own archive still lives separately in Dashboard and can be matched against them after a checkup.