最佳数据分析AI模型

使用MMLU、MATH-500和GPQA基准测试分数对AI模型的数据分析能力进行排名。找到数据科学、分析和洞察的最佳LLM。

Last updated: just now

第一名模型

o4 Mini

评分: 97.3

平均评分

69.8

所有排名模型

已排名模型

有基准测试数据

权重：MMLU (40%)MATH-500 (30%)GPQA (30%)

#	模型	提供商	评分	MMLU	MATH-500	GPQA
1	o4 MiniOpenAI	OpenAI	97.3	--	97.3	--
2	o3OpenAI	OpenAI	95.2	92.3	99	--
3	GPT-5.4OpenAI	OpenAI	94.6	94	95.5	--
4	R1 0528DeepSeek	DeepSeek	94.2	91.5	97.8	--
5	o1OpenAI	OpenAI	93.8	91.8	96.4	--
6	GPT-5.2OpenAI	OpenAI	93.7	93.5	94	--
7	R1DeepSeek	DeepSeek	93.6	90.8	97.3	--
8	Grok 4xAI	xAI	93	91.5	95	--
9	GPT-5OpenAI	OpenAI	92.8	93	92.5	--
10	Gemini 2.5 ProGoogle	Google	92.7	90.8	95.2	--
11	o3 MiniOpenAI	OpenAI	91.6	86.9	97.9	--
12	Claude Opus 4.6Anthropic	Anthropic	91.4	92.1	90.5	--
13	DeepSeek V3 0324DeepSeek	DeepSeek	90.4	89.2	92	--
14	Claude Opus 4.5Anthropic	Anthropic	90	91.4	88.1	--
15	DeepSeek V3DeepSeek	DeepSeek	89.2	88.5	90.2	--
16	Claude Opus 4Anthropic	Anthropic	88.9	91	86	--
17	Claude Sonnet 4.6Anthropic	Anthropic	88.7	91.2	85.3	--
18	Gemini 3 Flash PreviewGoogle	Google	88	88	88	--
19	Claude Sonnet 4.5Anthropic	Anthropic	87.5	90.8	83	--
20	Grok 3xAI	xAI	87	88.5	85	--
21	Claude 3.7 SonnetAnthropic	Anthropic	86.8	90.2	82.2	--
22	Claude Sonnet 4Anthropic	Anthropic	86	89.5	81.4	--
23	Gemini 2.5 FlashGoogle	Google	85.8	85.8	85.8	--
24	Llama 4 MaverickMeta	Meta	85	88	81	--
25	GPT-4.1OpenAI	OpenAI	84.6	89.2	78.5	--
26	Claude 3.5 SonnetAnthropic	Anthropic	84.2	88.7	78.3	--
27	GPT-4oOpenAI	OpenAI	83.5	88.7	76.6	--
28	Llama 3.3 70B InstructMeta	Meta	82.3	86.3	77	--
29	Gemini 2.0 FlashGoogle	Google	82.1	76.4	89.7	--
30	Mistral LargeMistral AI	Mistral AI	81	84.7	76	--
31	GPT-4 TurboOpenAI	OpenAI	80.5	86.5	72.6	--
32	Claude Haiku 4.5Anthropic	Anthropic	79.4	84.5	72.5	--
33	Llama 3.1 70B InstructMeta	Meta	78.3	86	68	--
34	GPT-4o-miniOpenAI	OpenAI	76.9	82	70.2	--
35	Claude 3.5 HaikuAnthropic	Anthropic	75.9	80.9	69.2	--
36	Gemma 2 27BGoogle	Google	75.2	75.2	--	--
37	Phi 4Microsoft	Microsoft	64.3	84.8	80.4	20.8
38	Qwen2.5 72B InstructAlibaba	Alibaba	16.7	--	--	16.7
39	Qwen2.5 Coder 32B InstructAlibaba	Alibaba	13.2	--	--	13.2
40	Gemma 2 9BGoogle	Google	9.7	--	--	9.7
41	Command R7B (12-2024)Cohere	Cohere	7.8	--	--	7.8
42	Llama 3.1 8B InstructMeta	Meta	7.4	--	--	7.4
43	Llama 3.2 3B InstructMeta	Meta	6.2	--	--	6.2
44	Qwen2.5 Coder 7B InstructAlibaba	Alibaba	5.8	--	--	5.8
45	Qwen2.5 7B InstructAlibaba	Alibaba	5.5	--	--	5.5
46	Llama 3 8B InstructMeta	Meta	2.1	--	--	2.1
47	QwQ 32BAlibaba	Alibaba	1.3	--	--	1.3