Horangi 4 - Ko-TruthfulQA: leaderboard
A Korean multiple-choice version of TruthfulQA, as sampled by the Horangi Korean LLM Leaderboard 4 (W&B Korea). Accuracy (%), higher is better.
Metric: Accuracy (%). Source: horangi.ai. Saturation forecast: Estimated already saturated. 105 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) (High) | 98 | #64 (Gemini 3.1 Pro (Preview)) |
| 2 | Claude Fable 5 (High) | 98 | #20 (Claude Fable 5) |
| 3 | GPT-5.5 (xHigh) | 96 | #33 (GPT-5.5) |
| 4 | Claude Opus 4.8 (High) | 96 | #39 (Claude Opus 4.8) |
| 5 | Claude Sonnet 4.5 (High) | 96 | #148 (Claude Sonnet 4.5) |
| 6 | Qwen 3.7 Max | 95 | #85 |
| 7 | Grok 4.5 | 95 | #54 |
| 8 | Claude Sonnet 5 (High) | 95 | #65 (Claude Sonnet 5) |
| 9 | Claude Opus 4.5 (20251101) (High) | 95 | #81 (Claude Opus 4.5 (20251101)) |
| 10 | Gemini 3.5 Flash | 93 | #66 |
| 11 | Gemini 3 Pro (Preview) (High) | 93 | #71 (Gemini 3 Pro (Preview)) |
| 12 | Gemini 3 Flash (Preview) | 92 | #90 |
| 13 | Grok 4.20 | 92 | #179 |
| 14 | GPT-5.6 Terra (Max) | 92 | #40 (GPT-5.6 Terra) |
| 15 | Claude Opus 4.6 (High) | 92 | #67 (Claude Opus 4.6) |
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-truthfulqa · How It Works · Data refreshed daily, snapshot 2026-11-05.