评分方法

我们的综合排名系统将多个基准测试合并为一个可比较的分数。以下是其工作原理。

Composite Score

The composite score is a weighted average of all tracked benchmark results for each model.

Benchmark Categories

• Reasoning (35%) — MMLU, GPQA, AIME, MATH
• Coding (30%) — HumanEval, SWE-Bench, LiveCodeBench
• Chat (20%) — Arena ELO, IFEval
• Multimodal (15%) — MMMU