Gaokao 2026
Category
—
Score type
percent
Scale
relative
higher is better
Models tested
7
What is Gaokao 2026?
Gaokao 2026 uses China's national college entrance examination (Gaokao) as a benchmark. The exam covers Chinese, mathematics, English and science subjects and is extremely difficult and broad. The score shows how well a model masters a realistic, very demanding curriculum-based test.
What does it test?
Gaokao 2026 measures the abilities the benchmark was built to measure. In detail, it checks:
- curriculum knowledge across many Chinese school subjects
- Chinese-language reading comprehension and essay-level content
- math and science problem solving under exam conditions
- broad coverage rather than one specialised skill
How to read the score
The score is accuracy: the percentage of tasks answered or solved correctly. 100% means every task was solved perfectly; higher is better.
Frequently asked questions
What is the Gaokao 2026 benchmark?
Gaokao 2026 uses China's national college entrance examination (Gaokao) as a benchmark. The exam covers Chinese, mathematics, English and science subjects and is extremely difficult and broad. The score shows how well a model masters a realistic, very demanding curriculum-based test.
What does Gaokao 2026 test?
Gaokao 2026 mainly measures the abilities the benchmark was built to measure. Concretely, it checks: curriculum knowledge across many Chinese school subjects; Chinese-language reading comprehension and essay-level content; math and science problem solving under exam conditions.
How is the Gaokao 2026 score calculated?
The score is accuracy: the percentage of tasks answered or solved correctly. 100% means every task was solved perfectly; higher is better.
Where do I get Gaokao 2026 data and scores?
The official benchmark page/reference is linked on this page. Every score on Open LLM Radar is also available as open data: download the scores JSON (http://open-llm-radar-api:8000/api/v1/benchmarks/gaokao-2026/scores) or pull it directly from the Open LLM Radar API with any HTTP client.
Benchmark data & downloads
Where to get, inspect and reuse the data behind this benchmark.
Model scores 7 models tested
| # | Model | Score | Date | Source | Verified |
|---|---|---|---|---|---|
| 1 |
Spark-X2.5-4B
SparkLLM (XHToken)
|
133.4%
|
— | Report | Verified |
| 2 |
Gemma4-12B
Google
|
130.6%
|
— | Report | Verified |
| 3 |
Qwen3.5-4B
Qwen
|
130.3%
|
— | Report | Verified |
| 4 |
Spark-X2.5-1.7B
SparkLLM (XHToken)
|
114.8%
|
— | Report | Verified |
| 5 |
Gemma4-E4B
Google
|
102.4%
|
— | Report | Verified |
| 6 |
Qwen3.5-2B
Qwen Team (Alibaba Cloud)
|
94%
|
— | Report | Verified |
| 7 |
Gemma4-E2B
Google
|
81.8%
|
— | Report | Verified |