KRETA - System 2: leaderboard

The 1,151 KRETA items that need domain knowledge or multi-step reasoning over the Korean text in the image, including adapted CSAT science and history questions; the score is the share answered correctly.

Metric: Accuracy on 1,151 System 2 (knowledge and multi-step reasoning) items (%). Source: github.com. Saturation forecast: Estimated already saturated. 17 models tracked. Updated 17 Nov 2026.

Top models

#ModelScoreOverall rank
1GPT-4o (2024-11-20)70.5#864
2Claude 3.5 Sonnet (20241022)64.5#648
3A.X-4.0-VL-Light56.5#2575
4VARCO-VISION-2.0-14B53.1#1794
5kanana-1.5-v-3B-instruct51.4#3953
6VARCO-VISION-14B49.3#1854
7InternVL2.5-8B47.3#2401
8InternVL2.5-4B45.9#2559
9Ovis1.6-Gemma2-9B45.4#2365
10Qwen 2.5 VL 3B Instruct43.9#2648
11MiniCPM-o-2.639.9#2417
12Ovis1.6-Llama3.2-3B39.1#3750
13Qwen 2.5 VL 7B Instruct36.1#1893
14DeepSeek-VL2-Small36.1#2502
15DeepSeek-VL2-Tiny34#6676

No result here: #1 Claude Fable 5.2, #2 GPT-6.2 Astra, #3 Gemini 4 Carbon.

Interactive version: theaggregate.ai/benchmark?slug=kreta-system-2 · How It Works · Data refreshed daily, snapshot 2026-11-17.