Korean CSAT 2026 (Easy Mode) - English: leaderboard
2026 Korean CSAT (college entrance exam, November 2025) solved by LLM configurations via official APIs without tools (hehee9/CSAT-benchmark), in Easy mode: one request per question, with OCR text, LaTeX and figures. English section (45 questions, listening replaced by transcripts); points out of 100, higher is better. The top of this board is close to saturated.
Metric: Points (out of 100). Source: hehee9.github.io. Saturation forecast: Estimated already saturated. 151 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.5 | 100 | #87 |
| 2 | GPT-5.6 Sol (Max) | 100 | #22 (GPT-5.6 Sol) |
| 3 | Gemini 3.8 Flash (High) | 100 | #30 (Gemini 3.8 Flash) |
| 4 | Gemini 3.1 Pro (Preview) (High) | 100 | #64 (Gemini 3.1 Pro (Preview)) |
| 5 | GPT-5.5 (xHigh) | 100 | #33 (GPT-5.5) |
| 6 | GPT-5.5 (High) | 100 | #33 (GPT-5.5) |
| 7 | GPT-5.6 Terra (Max) | 100 | #40 (GPT-5.6 Terra) |
| 8 | Gemini 3.5 Flash (High) | 100 | #66 (Gemini 3.5 Flash) |
| 9 | GPT-5.6 Sol (High) | 100 | #22 (GPT-5.6 Sol) |
| 10 | GPT-5 Mini (High) | 100 | #193 (GPT-5 Mini) |
| 11 | Gemini 3.7 Flash (High) | 100 | #24 (Gemini 3.7 Flash) |
| 12 | GPT-5.1 (High) | 100 | #145 (GPT-5.1) |
| 13 | DeepSeek V4 Pro (0813) (Max) | 100 | #58 (DeepSeek V4 Pro (0813)) |
| 14 | GPT-6 Astra (High) | 100 | #6 (GPT-6 Astra) |
| 15 | GPT-5.4 (xHigh) | 100 | #83 (GPT-5.4) |
No result here: #3 Claude Opus 5.5, #7 GPT-6.1 Sol, #9 Gemini 4 Argon.
Interactive version: theaggregate.ai/benchmark?slug=korean-csat-2026-easy-mode-english · How It Works · Data refreshed daily, snapshot 2026-11-05.