KOFFVQA - Commonsense Reasoning: leaderboard
KOFFVQA subscore for commonsense reasoning grounded in Korean visual question answering.
Metric: Score (%). Source: huggingface.co. Saturation forecast: Estimated already saturated. 82 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | O3 (2025-04-16) | 96.22 | #121 |
| 2 | GPT-4.5 (Preview) | 93.56 | #244 |
| 3 | O1 (2024-12-17) | 92.67 | #149 |
| 4 | Gemini 2.5 Flash (Preview 04-17) | 92.67 | #237 |
| 5 | Gemini 2.5 Flash (Preview 05-20) | 92 | #216 |
| 6 | GPT-4.1 | 91.56 | #248 |
| 7 | GPT-4.1 Mini | 89.56 | #355 |
| 8 | Gemini 2.5 Pro (Preview 05-06) | 89.33 | #147 |
| 9 | Gemini 2.5 Pro (Preview 03-25) | 89.11 | #123 |
| 10 | Claude 3.5 Sonnet (20241022) | 88.89 | #291 |
| 11 | GPT-4o (2024-08-06) | 87.56 | #336 |
| 12 | Gemini 2.0 Pro (Preview 02-05) | 87.11 | #307 |
| 13 | Claude 3.7 Sonnet (20250219) | 86.89 | #201 |
| 14 | GPT-4o (2024-11-20) | 85.56 | #381 |
| 15 | Qwen 2.5 VL 72B Instruct | 85.33 | #377 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=koffvqa-commonsense-reasoning · How It Works · Data refreshed daily, snapshot 2026-10-20.