Horangi 4 - Ko-Moral: leaderboard
Ko-Moral, Korean moral-judgment questions, as sampled by the Horangi Korean LLM Leaderboard 4 (W&B Korea). Accuracy (%), higher is better.
Metric: Accuracy (%). Source: horangi.ai. Saturation forecast: Around February 2027. 105 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Flash (Preview) | 90 | #90 |
| 2 | GPT-5.6 Sol (Max) | 88 | #22 (GPT-5.6 Sol) |
| 3 | GPT-5.5 (xHigh) | 88 | #33 (GPT-5.5) |
| 4 | Gemini 3.1 Pro (Preview) (High) | 87 | #64 (Gemini 3.1 Pro (Preview)) |
| 5 | GPT-5 | 86 | #105 |
| 6 | Gemini 3 Pro (Preview) (High) | 85 | #71 (Gemini 3 Pro (Preview)) |
| 7 | Claude Sonnet 4.5 (High) | 84 | #148 (Claude Sonnet 4.5) |
| 8 | GLM-5.2 (High) | 84 | #80 (GLM-5.2) |
| 9 | GPT-5.6 Luna (Max) | 83 | #79 (GPT-5.6 Luna) |
| 10 | Claude Opus 4.8 (High) | 83 | #39 (Claude Opus 4.8) |
| 11 | Claude Opus 4.1 (20250805) | 83 | #139 |
| 12 | GPT-5.2 (2025-12-11) (xHigh) | 83 | |
| 13 | GPT-4o (2024-11-20) | 82 | #392 |
| 14 | Claude Opus 4.6 (High) | 82 | #67 (Claude Opus 4.6) |
| 15 | Claude Fable 5 (High) | 82 | #20 (Claude Fable 5) |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=horangi-4-ko-moral · How It Works · Data refreshed daily, snapshot 2026-11-05.