PCT Political Inconsistency (CAIS)
Safety · 2026-05-21
Polarized Contrastive Pairs evaluates covert political manipulation by comparing model responses to matched prompts about opposing political subjects. SAFE reports Political Inconsistency, the complement of Political Consistency across 1,000 paired evaluations, so lower scores indicate more symmetric sentiment and helpfulness.
Top models (lower is better)
| Model | Score |
|---|---|
| Muse Spark 1.1 | 27.9 |
| Claude Opus 5.5 | 29.1 |
| Grok 4 Fast | 30.8 |
| Grok 4.1 Fast | 32.5 |
| Muse Wonder | 33.3 |
| Claude Fable 5.1 | 33.4 |
| GPT-5 | 35.3 |
| Claude Opus 5 | 36.7 |
| Gemini 2.5 Pro | 37.3 |
| GPT-6.1 Sol | 37.3 |
| GPT-6 Astra | 37.4 |
| Gemini 4 Argon | 38.7 |
| o3 | 39.0 |
| Claude Fable 5 | 39.8 |
| Grok 4.20 | 39.9 |
| GPT-5.5 | 40.0 |
| Muse Spark 1.3 | 40.6 |
| Muse Spark Open | 41.3 |
| GPT-5.6 Sol | 41.7 |
| Gemini 3.1 Pro Preview | 42.2 |
| Kimi K2.6 | 43.6 |
| DeepSeek-V4-Pro | 44.0 |
| GPT-5.6 Luna | 44.7 |
| Gemini 3.7 Flash | 45.1 |
| GPT-5.6 Terra | 46.2 |
| Gemini 3.8 Flash | 46.3 |
| GLM-5.3 | 46.8 |
| Kimi K3 | 47.6 |
| GPT-5.1 | 47.9 |
| Opus 4.7 | 48.2 |
| Opus 4.5 | 49.4 |
| Grok 4 | 49.5 |
| GLM-5.2 | 50.0 |
| Grok 4.5 | 50.2 |
| Grok 4.3 | 51.6 |
| Opus 4.6 | 51.7 |
| Gemini 3.1 Flash-Lite Preview | 51.9 |
| Gemini 3.5 Flash | 53.4 |
| Opus 4.8 | 53.8 |
| Grok 4.6 | 55.4 |