Gaokao-Bench: leaderboard
Gaokao-Bench: Evaluates broad language-model knowledge, reasoning, commonsense, instruction following, or exam-style accuracy.
Metric: Objective Overall (%). Source: github.com. Source: benchmarklist.com. Saturation forecast: Estimated already saturated. 13 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4 (0314) | 72.2 | #574 |
| 2 | GPT-4 (0613) | 71.6 | #491 |
| 3 | Gemini 1.0 Pro | 57.9 | #933 |
| 4 | Baichuan2-13B-Chat | 43.9 | #1255 |
| 5 | ChatGLM2 6B | 42.7 | #1498 |
| 6 | Baichuan2-7B-Chat | 40.5 | #1420 |
| 7 | LLaMA-7B | 21.1 | #1564 |
| 8 | Vicuna-7B | 21 | #1542 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=gaokao-bench · How It Works · Data refreshed daily, snapshot 2026-10-20.