Polar (South Korea) - Sociocultural Axis: leaderboard
The sociocultural axis of Polar (2026) on the South Korean set in Korean: an ICAT score from 0 to 100 combining language-modelling competence with balance between progressive and conservative positions, from option likelihoods. Higher is better.
Metric: Sociocultural-axis ICAT (0-100), the StereoSet idealized context association score adapted to political bias: the language-model score LMS (share of items where the unrelated continuation is not chosen) times the neutrality score NS (how close the preference between the left or progressive and the right or conservative continuation is to balance, averaged over issue categories); options chosen by length-normalized log-likelihood in LM Evaluation Harness; higher means more balanced and competent; South Korean political statements in Korean. Source: arxiv.org. Saturation forecast: Estimated already saturated. 38 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Llama 3.1 70B | 96.23 | #1490 |
| 2 | Mistral Small 3 | 93.34 | #1777 |
| 3 | Qwen3-14B-Base | 93.23 | #1504 |
| 4 | Mistral-Small-24B-Base-2501 | 92.38 | #1450 |
| 5 | Qwen 3 14B | 92.13 | #1233 |
| 6 | Midm-2.0-Mini-Instruct | 91.96 | #4735 |
| 7 | Llama 3.1 70B Instruct | 91.86 | #1546 |
| 8 | Qwen3-8B-Base | 90.48 | #1721 |
| 9 | A.X 4.0 | 90.22 | #863 |
| 10 | Llama 3.1 8B | 90.19 | #4861 |
| 11 | A.X-4.0-Light | 89.75 | #2254 |
| 12 | Qwen3-4B-Base | 89.31 | #2515 |
| 13 | Midm-2.0-Base-Instruct | 89.09 | #1921 |
| 14 | Qwen 3 8B | 89.06 | #1830 |
| 15 | Qwen 3 32B | 89.04 | #960 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=polar-south-korea-sociocultural-axis · How It Works · Data refreshed daily, snapshot 2026-11-13.