基准测试

NL2Repo-Bench

Natural-language-to-repository generation.

分数历史 →

模型分数

#	模型	分数
1	Claude Opus 5	75.3%
2	DeepSeek V4.1-Flash 1017	68.2%
3	DeepSeek V4.1-Flash 0910	65.4%
4	DeepSeek V4-Pro 0813	61.5%
5	GLM 5.3	58%
6	Kimi K3	58%
7	GPT 5.6-Sol	56.8%
8	DeepSeek V4-Flash 0731	54.2%
9	MiniMax M3	42.1%