What is the Artificial Analysis Intelligence Index?

The Artificial Analysis Intelligence Index v4.0 is a composite benchmark that aggregates performance across 10 evaluations spanning mathematics, science, coding, agentic tasks, and reasoning to measure AI capabilities holistically.

How is the Intelligence Index calculated?

The index aggregates scores from 10 benchmarks: GDPval-AA, τ²-Bench, Terminal-Bench Hard, SciCode, AA-LCR, AA-Omniscience, IFBench, Humanity's Last Exam, GPQA Diamond, and CritPt. All tests are independently run on standardized hardware.

How does this differ from LMArena?

LMArena uses crowdsourced user votes (Elo ratings) reflecting subjective preferences. The AA Intelligence Index uses standardized automated benchmarks with objective scoring across specific technical domains.

Where can I find the original data?

The original leaderboard is available at artificialanalysis.ai/leaderboards/models and the methodology at artificialanalysis.ai/evaluations/artificial-analysis-intelligence-index.

Artificial Analysis Intelligence Index AI模型智能指数排行榜

Name: Artificial Analysis Intelligence Index AI模型智能指数排行榜
Creator: DataLearner
License: https://creativecommons.org/licenses/by/4.0/

Artificial Analysis Intelligence Index v4.0 综合了10项权威评测基准（GDPval-AA、Terminal-Bench、GPQA Diamond、SciCode等），从数学、科学、编程、推理等多维度对AI模型进行全面评估和排名。

榜首模型

Claude Opus 5 (max)

最高得分

模型数量

251

数据版本

2026年08月02日

数据来源: Artificial Analysis

来源：全部国产模型

榜单历史快照月份:

排名总表

排名	模型名称	智能指数	机构
	Claude Opus 5 (max)Anthropic	61	Anthropic
	Claude Opus 5 (xhigh)Anthropic	60	Anthropic
	Claude Fable 5Anthropic	60	Anthropic
4	GPT-5.6 Sol (max)OpenAI	59	OpenAI
5	Claude Opus 5 (high)Anthropic	59	Anthropic
6	GPT-5.6 Sol (xhigh)OpenAI	58	OpenAI
7	Kimi K3 (max)Kimi	57	Kimi
8	Claude Opus 5 (medium)Anthropic	56	Anthropic
9	GPT-5.6 Sol (high)OpenAI	56	OpenAI
10	GPT-5.6 Terra (max)OpenAI	55	OpenAI
11	Grok 4.5 (high)xAI	54	xAI
12	GPT-5.6 Sol (medium)OpenAI	54	OpenAI
13	Claude Sonnet 5 (max)Anthropic	53	Anthropic
14	GPT-5.6 Terra (xhigh)OpenAI	52	OpenAI
15	GPT-5.6 Luna (max)OpenAI	51	OpenAI
16	GLM-5.2 (max)智谱AI	51	智谱AI
17	Muse Spark 1.1 (xhigh)Facebook AI研究实验室	51	Facebook AI研究实验室
18	Claude Opus 5 (low)Anthropic	51	Anthropic
19	Gemini 3.5 FlashGoogle Deep Mind	50	Google Deep Mind
20	Gemini 3.6 FlashGoogle	50	Google
21	DeepSeek V4 Flash 0731 (max)DeepSeek	50	DeepSeek
22	GPT-5.6 Sol (low)OpenAI	49	OpenAI
23	GPT-5.6 Luna (xhigh)OpenAI	49	OpenAI
24	GPT-5.6 Terra (high)OpenAI	49	OpenAI
25	Kimi K3 (low)Kimi	47	Kimi
26	Gemini 3.1 Pro PreviewGoogle Deep Mind	46	Google Deep Mind
27	GPT-5.6 Luna (high)OpenAI	46	OpenAI
28	Qwen3.7 Max阿里巴巴	46	阿里巴巴
29	GPT-5.6 Terra (medium)OpenAI	46	OpenAI
30	Gemini 3.5 Flash (medium)Google Deep Mind	45	Google Deep Mind
31	MiniMax M3MiniMaxAI	44	MiniMaxAI
32	DeepSeek-V4-Pro (max)DeepSeek-AI	44	DeepSeek-AI
33	GPT-5.3 Codex (xhigh)OpenAI	44	OpenAI
34	Motif 3 (Beta)Motif Technologies	44	Motif Technologies
35	DeepSeek-V4-Pro (high)DeepSeek-AI	43	DeepSeek-AI
36	Muse SparkFacebook AI研究实验室	43	Facebook AI研究实验室
37	MiMo-V2.5-ProXiaomi	42	Xiaomi
38	Kimi K2.7 CodeMoonshot AI	42	Moonshot AI
39	Claude Sonnet 5 (non-reasoning)Anthropic	42	Anthropic
40	Hy3腾讯AI实验室	41	腾讯AI实验室
41	GPT-5.6 Sol (non-reasoning)OpenAI	41	OpenAI
42	Nex-N2-ProNex AGI	41	Nex AGI
43	InklingThinking Machines Lab	41	Thinking Machines Lab
44	GPT-5.6 Terra (low)OpenAI	40	OpenAI
45	Inkling SmallThinking Machines	40	Thinking Machines
46	Qwen 3.6 Plus Preview阿里巴巴	40	阿里巴巴
47	Qwen3.7-Plus阿里巴巴	39	阿里巴巴
48	JT-4.1 Flash 236B A21BChina Mobile	39	China Mobile
49	Agnes 2.5 Pro AlphaSapiens AI	39	Sapiens AI
50	GPT-5.6 Luna (medium)OpenAI	38	OpenAI
51	Nemotron 3 UltraNVIDIA	38	NVIDIA
52	MiMo-V2.5Xiaomi	37	Xiaomi
53	Qwen3.6-27B阿里巴巴	37	阿里巴巴
54	Gemini 3.5 Flash-LiteGoogle	36	Google
55	Nova 2 Omni（Preview）亚马逊	36	亚马逊
56	Grok 4.3 Beta (medium)xAI	36	xAI
57	Grok 4.3 Beta (low)xAI	35	xAI
58	MiMo-V2-OmniXiaomi	35	Xiaomi
59	Gemini 3.5 Flash (minimal)Google Deep Mind	35	Google Deep Mind
60	Kimi K2.6Moonshot AI	35	Moonshot AI
61	Claude Sonnet 4.6 (Non-reasoning, Low Effort)Anthropic	34	Anthropic
62	GLM-5.2智谱AI	34	智谱AI
63	GPT-5.6 Terra (non-reasoning)OpenAI	34	OpenAI
64	KAT-Coder-Pro V2KwaiKAT	34	KwaiKAT
65	Qwen3.5-397B-A17B阿里巴巴	34	阿里巴巴
66	Hy3腾讯AI实验室	34	腾讯AI实验室
67	LongCat 2.0LongCat	33	LongCat
68	GPT-5.6 Luna (low)OpenAI	33	OpenAI
69	Gemini 2.0 Flash ExperimentalDeepMind	33	DeepMind
70	Qwen3.5-122B-A10B阿里巴巴	32	阿里巴巴
71	Qwen3.5-397B-A17B阿里巴巴	32	阿里巴巴
72	Qwen3.6-35B-A3B阿里巴巴	32	阿里巴巴
73	DeepSeek-V4-ProDeepSeek-AI	31	DeepSeek-AI
74	Qwen3.5-Omni-Plus阿里巴巴	31	阿里巴巴
75	Ring-2.6-1TInclusionAI	31	InclusionAI
76	Qwen3.6-27B阿里巴巴	30	阿里巴巴
77	OpenAI o3OpenAI	30	OpenAI
78	Step 3.7 FlashStepFunAI	30	StepFunAI
79	Mistral Medium 3.5MistralAI	30	MistralAI
80	Haiku 4.5Anthropic	30	Anthropic
81	Gemma 4 31BDeepMind	29	DeepMind
82	GPT-5.5 InstantOpenAI	29	OpenAI
83	DeepSeek-V4-FlashDeepSeek-AI	29	DeepSeek-AI
84	JT-35B-FlashChina Mobile	28	China Mobile
85	KAT-Coder-Pro V1KwaiKAT	28	KwaiKAT
86	MiMo-V2.5-ProXiaomi	28	Xiaomi
87	Qwen3.5-122B-A10B阿里巴巴	28	阿里巴巴
88	GPT-5.6 Luna (non-reasoning)OpenAI	27	OpenAI
89	Hy3腾讯AI实验室	26	腾讯AI实验室
90	Ling-2.6-1TInclusionAI	26	InclusionAI
91	Doubao Seed CodeByteDance Seed	26	ByteDance Seed
92	Gemini 2.5 ProGoogle Deep Mind	26	Google Deep Mind
93	Gemma 4 26B A4BDeepMind	26	DeepMind
94	NVIDIA Nemotron 3 SuperNVIDIA	25	NVIDIA
95	Gemini 3.1 Flash-LiteGoogle Deep Mind	25	Google Deep Mind
96	Grok 4.3 Beta (non-reasoning)xAI	25	xAI
97	MiMo-V2-FlashXiaomi	25	Xiaomi
98	Qwen3.6-35B-A3B阿里巴巴	24	阿里巴巴
99	Qwen3.5-35B-A3B阿里巴巴	24	阿里巴巴
100	GPT OSS 120B (high)OpenAI	24	OpenAI
101	Haiku 4.5Anthropic	24	Anthropic
102	C4AI Command A (202503)CohereAI	23	CohereAI
103	K-EXAONELG AI Research	22	LG AI Research
104	ERNIE 5.0 Thinking Preview百度	22	百度
105	Gemma 4 12BGoogle	22	Google
106	Gemma 4 31BDeepMind	22	DeepMind
107	Nova 2 Pro（Preview） (medium)亚马逊	22	亚马逊
108	Qwen3.5-9B阿里巴巴	21	阿里巴巴
109	Mercury 2Inception	21	Inception
110	Qwen3-Coder-Next阿里巴巴	21	阿里巴巴
111	Nova 2 Omni（Preview） (medium)亚马逊	21	亚马逊
112	Apriel-v1.6-15B-ThinkerServiceNow	21	ServiceNow
113	Qwen3.5-9B阿里巴巴	20	阿里巴巴
114	EXAONE 4.5 33BLG AI Research	20	LG AI Research
115	Gemma 4 26B A4BDeepMind	20	DeepMind
116	Qwen3.5 4BAlibaba	20	Alibaba
117	North Mini CodeCohere	20	Cohere
118	Nova 2 Pro（Preview） (low)亚马逊	20	亚马逊
119	Mistral Small 4Mistral	20	Mistral
120	Devstral 2Mistral	19	Mistral
121	Nova 2 Lite (medium)亚马逊	19	亚马逊
122	Qwen3.5-Omni-Flash阿里巴巴	19	阿里巴巴
123	JT-MINIChina Mobile	19	China Mobile
124	Nova 2 Lite (high)亚马逊	18	亚马逊
125	Trinity Large ThinkingArcee AI	18	Arcee AI
126	Magistral Medium 1.2Mistral	18	Mistral
127	Nova 2 Lite (low)亚马逊	18	亚马逊
128	HyperNova 60B 2605Multiverse Computing	18	Multiverse Computing
129	Nemotron Cascade 2 30B A3BNVIDIA	18	NVIDIA
130	Devstral Small 2Mistral	17	Mistral
131	K2 Think V2MBZUAI	17	MBZUAI
132	LongCat Flash LiteLongCat	17	LongCat
133	HyperCLOVA X SEED Think (32B)Naver	17	Naver
134	K-EXAONELG AI Research	17	LG AI Research
135	Qwen3-Next阿里巴巴	17	阿里巴巴
136	Nova 2 Omni（Preview） (low)亚马逊	17	亚马逊
137	Mi:dm K 2.5 ProKorea Telecom	16	Korea Telecom
138	G9v3-3BAI9Stars	16	AI9Stars
139	Qwen3.5 4BAlibaba	16	Alibaba
140	Mistral Large 3MistralAI	16	MistralAI
141	INTELLECT-3Prime Intellect	16	Prime Intellect
142	Solar Open 100BUpstage	15	Upstage
143	Qwen3-Omni-30B-A3B (reasoning)阿里巴巴	15	阿里巴巴
144	GPT OSS 120B (low)OpenAI	15	OpenAI
145	GPT OSS 20B (high)OpenAI	15	OpenAI
146	Nova 2 Pro（Preview）亚马逊	14	亚马逊
147	GPT OSS 20B (low)OpenAI	14	OpenAI
148	Llama 4 MaverickFacebook AI研究实验室	14	Facebook AI研究实验室
149	K2-V2 (high)MBZUAI	14	MBZUAI
150	NVIDIA Nemotron 3 NanoNVIDIA	14	NVIDIA
151	Solar Pro 3Upstage	14	Upstage
152	Ling 2.6 FlashInclusionAI	14	InclusionAI
153	Qwen3-Next阿里巴巴	14	阿里巴巴
154	DiffusionGemma 26B A4BGoogle	13	Google
155	Gemma 4 12B (Non-reasoning)Google	13	Google
156	Motif-2-12.7BMotif Technologies	13	Motif Technologies
157	Nova PremierAmazon	13	Amazon
158	K2-V2 (medium)MBZUAI	12	MBZUAI
159	Llama Nemotron Super 49B v1.5Meta	12	Meta
160	Mistral Small 4Mistral	12	Mistral
161	Tri-21B-ThinkTrillion Labs	12	Trillion Labs
162	Gemma 4 E4BDeepMind	12	DeepMind
163	MiniCPM5-1BOpenBMB	12	OpenBMB
164	Sarvam 105B (high)Sarvam	12	Sarvam
165	Nova 2 Lite亚马逊	12	亚马逊
166	Celeris-1Celeris	12	Celeris
167	MiniCPM5-1BOpenBMB	12	OpenBMB
168	Magistral Small 1.2Mistral	11	Mistral
169	Nanbeige4.1-3BNanbeige	11	Nanbeige
170	Ministral 3 14BMistralAI	11	MistralAI
171	EXAONE 4.0 32BLG AI Research	11	LG AI Research
172	Nova 2 Omni（Preview）亚马逊	11	亚马逊
173	Llama 4 ScoutFacebook AI研究实验室	10	Facebook AI研究实验室
174	Hermes 4 70BNous Research	10	Nous Research
175	Falcon-H1R-7BTII UAE	10	TII UAE
176	Gemma 4 E2BDeepMind	10	DeepMind
177	Qwen3-Omni-30B-A3B阿里巴巴	10	阿里巴巴
178	Step3 VL 10BStepFun	9	StepFun
179	Llama3.3-70B-InstructFacebook AI研究实验室	9	Facebook AI研究实验室
180	Llama Nemotron UltraNVIDIA	9	NVIDIA
181	ERNIE-4.5-300B-A47B百度	9	百度
182	Hermes 4 405BNous Research	9	Nous Research
183	NVIDIA Nemotron Nano 12B v2 VLNVIDIA	9	NVIDIA
184	Ministral 3 8BMistralAI	9	MistralAI
185	Gemma 4 E4BDeepMind	9	DeepMind
186	Granite 4.1 30BIBM	9	IBM
187	NVIDIA Nemotron Nano 9B V2NVIDIA	9	NVIDIA
188	Hermes 4 405BNous Research	9	Nous Research
189	NVIDIA Nemotron 3 Nano 4BNVIDIA	9	NVIDIA
190	Llama Nemotron Super 49B v1.5Meta	9	Meta
191	K2-V2 (low)MBZUAI	9	MBZUAI
192	Kimi Linear 48B A3B InstructKimi	9	Kimi
193	Llama3.1-405BFacebook AI研究实验室	9	Facebook AI研究实验室
194	LFM2.5-8B-A1BLiquid AI	8	Liquid AI
195	Ring-flash-2.0InclusionAI	8	InclusionAI
196	Olmo 3.1 32B ThinkAI2	8	AI2
197	C4AI Command A (202503)CohereAI	8	CohereAI
198	Qwen3.5 2BAlibaba	8	Alibaba
199	Llama 3.1 Nemotron 70BNVIDIA	8	NVIDIA
200	NVIDIA Nemotron 3 NanoNVIDIA	7	NVIDIA
201	NVIDIA Nemotron Nano 9B V2NVIDIA	7	NVIDIA
202	Hermes 4 70BNous Research	7	Nous Research
203	Ministral 3 3BMistral	7	Mistral
204	Granite 4.1 8BIBM	7	IBM
205	Sarvam 30B (high)Sarvam	7	Sarvam
206	Olmo 3.1 32B InstructAI2	6	AI2
207	Gemma 4 E2BDeepMind	6	DeepMind
208	R1 1776Perplexity	6	Perplexity
209	Llama 3.2-Vision-90BFacebook AI研究实验室	6	Facebook AI研究实验室
210	Phi-4-mini-instruct (3.8B)Microsoft Azure	6	Microsoft Azure
211	EXAONE 4.0 32BLG AI Research	6	LG AI Research
212	Qwen3.5 2BAlibaba	6	Alibaba
213	Qwen3.5 0.8BAlibaba	5	Alibaba
214	DeepHermes 3 - Mistral 24BNous Research	5	Nous Research
215	Jamba 1.7 LargeAI21 Labs	5	AI21 Labs
216	Granite 4.0 H SmallIBM	5	IBM
217	Qwen3-Omni-30B-A3B阿里巴巴	5	阿里巴巴
218	LFM2 24B A2BLiquid AI	5	Liquid AI
219	Phi-4-reasoningMicrosoft Azure	5	Microsoft Azure
220	Amazon Nova Micro亚马逊	5	亚马逊
221	Granite 4.1 3BIBM	5	IBM
222	NVIDIA Nemotron Nano 12B v2 VLNVIDIA	5	NVIDIA
223	Phi-4-multimodal-instruct Microsoft Azure	5	Microsoft Azure
224	MiniCPM-V 4.6 1.3BOpenBMB	4	OpenBMB
225	Jamba Reasoning 3BAI21 Labs	4	AI21 Labs
226	Gemini 3.0 FlashGoogle Deep Mind	4	Google Deep Mind
227	Olmo 3 7B ThinkAI2	4	AI2
228	Molmo 7B-DAllen Institute for AI	4	Allen Institute for AI
229	Ling-mini-2.0InclusionAI	4	InclusionAI
230	Llama 3.2-Vision-11BFacebook AI研究实验室	3	Facebook AI研究实验室
231	Qwen3.5 0.8BAlibaba	3	Alibaba
232	Exaone 4.0 1.2BLG AI Research	3	LG AI Research
233	Olmo 3 7BAI2	3	AI2
234	Exaone 4.0 1.2BLG AI Research	3	LG AI Research
235	LFM2.5-1.2B-ThinkingLiquid AI	3	Liquid AI
236	Jamba 1.7 MiniAI21 Labs	3	AI21 Labs
237	LFM2 2.6BLiquid AI	3	Liquid AI
238	LFM2.5-1.2B-InstructLiquid AI	3	Liquid AI
239	Granite 4.0 H 1BIBM	3	IBM
240	Gemma 3-270MGoogle Deep Mind	2	Google Deep Mind
241	Apertus 70B InstructSwiss AI	2	Swiss AI
242	Granite 4.0 MicroIBM	2	IBM
243	DeepHermes 3 - Llama-3.1 8BNous Research	2	Nous Research
244	Granite 4.0 1BIBM	2	IBM
245	Molmo2-8BAI2	2	AI2
246	LFM2 8B A1BLiquid AI	2	Liquid AI
247	LFM2.5-VL-1.6BLiquid AI	1	Liquid AI
248	Granite 4.0 350MIBM	1	IBM
249	Tiny Aya GlobalCohere	1	Cohere
250	Apertus 8B InstructSwiss AI	1	Swiss AI
251	Granite 4.0 H 350MIBM	1	IBM

数据仅供参考，以官方来源为准。模型名称旁的链接可跳转到 DataLearner 模型详情页。

评测基准组成（Intelligence Index v4.0）

Intelligence Index 综合10项严格的评测基准，全面衡量AI模型能力，避免单一维度的过拟合。

GDPval-AA

智能体真实任务

τ²-Bench

智能体工具调用

Terminal-Bench

智能体编程

SciCode

编程能力

AA-LCR

长上下文推理

AA-Omniscience

知识与幻觉检测

IFBench

指令遵循

Humanity's Last Exam

推理与知识

GPQA Diamond

科学推理

CritPt

物理推理

常见问题 (FAQ)

什么是 Artificial Analysis Intelligence Index？▼

Artificial Analysis Intelligence Index v4.0 是一个综合评测指数，聚合了10项具有挑战性的评估——涵盖数学、科学、编程、智能体任务和推理——以全面衡量AI能力。它旨在防止单一维度的过拟合，提供一个统一分数来追踪模型进步。

智能指数是如何计算的？▼

该指数综合了10项评测的分数：GDPval-AA（智能体真实任务）、τ²-Bench（工具调用）、Terminal-Bench Hard（智能体编程）、SciCode（编程）、AA-LCR（长上下文推理）、AA-Omniscience（知识与幻觉检测）、IFBench（指令遵循）、Humanity's Last Exam（推理）、GPQA Diamond（科学推理）和 CritPt（物理推理）。所有测试由 Artificial Analysis 在标准化硬件上独立运行。

这与 LMArena 排行榜有什么区别？▼

LMArena 排名基于众包用户投票（盲测A/B对比的Elo评分），反映主观的人类偏好。而 Artificial Analysis Intelligence Index 使用标准化的自动评测基准进行客观评分，衡量特定领域的技术能力。两者各有价值——LMArena 捕捉真实用户体验，而 AA Intelligence Index 提供可复现的技术测量。

在哪里可以找到原始数据？▼

原始排行榜和详细方法论可在 artificialanalysis.ai 查看。Intelligence Index 的方法论详见 Intelligence Index 页面。