Benchmark

大语言模型综合能力榜

general text 多语言

大语言模型综合能力榜：多维度评测综合成绩。评测分数为事实数据，整理自公开评测体系，仅供学习研究参考。

Top 10

榜单概览

数据说明

分数为 0-100 百分制综合成绩（Elo 榜为相对评分，仅供同榜内比较）。评测分数属事实数据，整理自公开评测体系与厂商公开资料，部分分数为厂商自报口径，不同榜单的评测方法与数据集可能存在差异，请结合多个榜单交叉参考。

Ranking

模型排名

名次	模型	机构	主分	语言	知识	推理	数学	代码	智能体
1	GPT-5.4	OpenAI	67.3	80.2	93.7	64.4	72.1	63.4	52.8
2	DeepSeek V4 Pro	DeepSeek	65.1	70.1	92.1	60.0	74.8	59.4	55.3
3	Claude Opus 4.7	Anthropic	64.0	70.0	92.2	63.5	72.4	63.7	43.3
4	Doubao-Seed-2-0-Pro-260215 (high)	ByteDance	63.5	77.3	91.9	65.2	77.3	57.0	44.2
5	Kimi K2.6	Moonshot AI	63.4	75.0	91.5	62.2	76.4	62.4	39.3
6	Gemini 3.1 Pro Preview	Google	63.2	74.8	95.2	59.7	76.5	60.2	41.0
7	Qwen3.6 Max Preview	Alibaba Cloud / Qwen Team	60.8	79.5	89.6	59.2	73.4	61.6	30.5
8	DeepSeek V4 Flash	DeepSeek	60.4	66.2	88.6	57.2	69.2	56.4	46.7
9	Doubao-Seed-2-0-Lite-260215 (high)	ByteDance	60.4	74.4	88.9	59.5	72.4	54.6	42.4
10	GLM-5.1	Zhipu AI	59.0	71.9	89.9	54.3	67.9	55.7	39.9
11	Qwen3.5 397B-A17B	Alibaba Cloud / Qwen Team	57.5	71.0	91.2	54.5	70.9	57.4	28.7
12	Claude Sonnet 4.6	Anthropic	57.5	70.5	90.6	60.6	72.4	58.1	24.7
13	Hy3 preview	Tencent	56.5	74.4	88.2	58.5	74.5	51.9	28.7
14	Ring-2.5-1T	Ant Group	54.5	69.8	89.4	54.2	66.5	53.6	25.0
15	GPT-5.4 mini	OpenAI	54.4	73.0	91.5	55.9	66.1	52.9	22.0
16	MiniMax-M2.7	MiniMax	53.7	68.1	82.4	52.4	64.7	56.2	23.7
17	Grok-4.20-0309	xAI	53.5	73.9	92.9	55.3	72.5	53.3	13.3
18	Step 3.5 Flash	StepFun	52.9	64.4	86.8	55.6	68.7	54.9	18.0
19	Gemma 4 31B IT	Google	52.4	64.6	87.7	54.6	66.5	56.8	14.0
20	Qwen3.6 27B	Alibaba Cloud / Qwen Team	52.3	69.0	84.0	52.7	65.6	50.8	23.3
21	Gemini 3.1 Flash Lite Preview	Google	49.9	58.7	91.7	52.0	66.4	52.8	11.0
22	JT-MINI	China Mobile	39.9	60.8	80.4	34.1	41.5	38.5	13.3
23	Gemma 4 E4B IT	Google	35.4	55.8	76.4	30.6	36.8	38.9	1.3