Horangi 4 - Ko-AIME 2025: leaderboard
AIME 2025 competition problems translated into Korean, as run by the Horangi Korean LLM Leaderboard 4 (W&B Korea). Accuracy (%) of final answers, higher is better.
Metric: Accuracy (%). Source: horangi.ai. Saturation forecast: Estimated already saturated. 105 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Grok 4.20 | 100 | #179 |
| 2 | GPT-5.5 (xHigh) | 100 | #33 (GPT-5.5) |
| 3 | Grok 4.5 | 100 | #54 |
| 4 | Claude Opus 4.8 (High) | 100 | #39 (Claude Opus 4.8) |
| 5 | Claude Fable 5 (High) | 100 | #20 (Claude Fable 5) |
| 6 | GPT-5.2 (2025-12-11) (xHigh) | 100 | |
| 7 | GPT-5.4 (2026-03-05) (xHigh) | 100 | #63 (GPT-5.4 (2026-03-05)) |
| 8 | A.X K1 (Thinking) | 100 | #441 (A.X K1) |
| 9 | GPT-5.4 Nano (2026-03-17) (xHigh) | 100 | #285 (GPT-5.4 Nano (2026-03-17)) |
| 10 | Gemini 3 Flash (Preview) | 96.67 | #90 |
| 11 | GPT-5 Mini (2025-08-07) | 96.67 | #175 |
| 12 | Qwen 3.7 Max | 96.67 | #85 |
| 13 | GPT-5.6 Sol (Max) | 96.67 | #22 (GPT-5.6 Sol) |
| 14 | Gemini 3.1 Pro (Preview) (High) | 96.67 | #64 (Gemini 3.1 Pro (Preview)) |
| 15 | GPT-5.6 Luna (Max) | 96.67 | #79 (GPT-5.6 Luna) |
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-aime-2025 · How It Works · Data refreshed daily, snapshot 2026-11-05.