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 Fable 5.1

最高得分

模型数量

270

数据版本

2026年09月22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 Fable 5.1Anthropic	53	Anthropic
	Claude Fable 5.1Anthropic	53	Anthropic
	GPT-6 Astra (max)OpenAI	53	OpenAI
4	GPT-6 Astra (xhigh)OpenAI	52	OpenAI
5	Claude Fable 5.1Anthropic	51	Anthropic
6	GPT-6 Astra (high)OpenAI	51	OpenAI
7	Claude Opus 5 (max)Anthropic	51	Anthropic
8	Claude Opus 5 (xhigh)Anthropic	50	Anthropic
9	GPT-6 Astra (medium)OpenAI	50	OpenAI
10	Claude Fable 5.1Anthropic	49	Anthropic
11	Claude Opus 5 (high)Anthropic	48	Anthropic
12	Muse Spark 1.3 (max)Meta	48	Meta
13	GPT-5.6 Sol (max)OpenAI	47	OpenAI
14	Claude Fable 5.1Anthropic	47	Anthropic
15	Grok 4.7 (xhigh)SpaceXAI	46	SpaceXAI
16	Grok 4.7 (high)SpaceXAI	46	SpaceXAI
17	GPT-6 Astra (low)OpenAI	46	OpenAI
18	Qwen3.8 Max (0902)Alibaba	45	Alibaba
19	Muse Spark 1.3 (xhigh)Meta	45	Meta
20	Claude Opus 5 (medium)Anthropic	45	Anthropic
21	GLM-5.3 (max)智谱AI	45	智谱AI
22	Grok 4.6 (high)xAI	44	xAI
23	Grok 4.6 (xhigh)xAI	44	xAI
24	GPT-5.6 Sol (xhigh)OpenAI	44	OpenAI
25	Step 5 PreviewStepFun	44	StepFun
26	Kimi K3 (max)Moonshot AI	44	Moonshot AI
27	Grok 4.6 (medium)xAI	43	xAI
28	GPT-5.6 Sol (high)OpenAI	42	OpenAI
29	GPT-5.6 Terra (max)OpenAI	42	OpenAI
30	GLM-5.3-FlashZ AI	42	Z AI
31	Gemini 3.8 Flash (high)Google	41	Google
32	Qwen3.8 2.4T A95BAlibaba	40	Alibaba
33	Qwen3.8-Flash-NextAlibaba	40	Alibaba
34	Gemini 3.8 Flash (medium)Google	40	Google
35	DeepSeek V4.1 Flash (max)DeepSeek	39	DeepSeek
36	Claude Opus 5 (low)Anthropic	39	Anthropic
37	GPT-5.6 Sol (medium)OpenAI	39	OpenAI
38	Claude Sonnet 5 (max)Anthropic	38	Anthropic
39	GPT-5.6 Terra (xhigh)OpenAI	38	OpenAI
40	GPT-5.6 Luna (max)OpenAI	37	OpenAI
41	DeepSeek-V4-Pro (max)DeepSeek-AI	36	DeepSeek-AI
42	Agnes 3.0 FlashSapiens AI	36	Sapiens AI
43	Agnes 2.5 Pro BetaSapiens AI	35	Sapiens AI
44	Grok 4.6 (low)xAI	35	xAI
45	DeepSeek V4 Flash Vision (max)DeepSeek	35	DeepSeek
46	GPT-5.6 Luna (xhigh)OpenAI	35	OpenAI
47	Kimi K3 (low)Moonshot AI	34	Moonshot AI
48	Claude Sonnet 5 (xhigh)Anthropic	34	Anthropic
49	GPT-5.6 Terra (high)OpenAI	34	OpenAI
50	Qwen3.8-27B (xhigh)阿里巴巴	34	阿里巴巴
51	Motif 3Motif Technologies	34	Motif Technologies
52	GPT-5.6 Sol (low)OpenAI	33	OpenAI
53	Gemini 3.8 Flash (low)Google	33	Google
54	GPT-5.3 Codex (xhigh)OpenAI	33	OpenAI
55	Motif 3 (Beta)Motif Technologies	32	Motif Technologies
56	GPT-5.6 Luna (high)OpenAI	32	OpenAI
57	Claude Sonnet 5 (high)Anthropic	32	Anthropic
58	K2 Horizon 375B A23BMBZUAI Institute of Foundation Models	31	MBZUAI Institute of Foundation Models
59	Apodex 1.1Apodex	30	Apodex
60	GPT-5.6 Terra (medium)OpenAI	30	OpenAI
61	Gemini 3.1 Pro PreviewGoogle DeepMind	30	Google DeepMind
62	MiniMax M3MiniMaxAI	29	MiniMaxAI
63	GPT-5.6 Sol (non-reasoning)OpenAI	28	OpenAI
64	Nex-N2-ProNex AGI	28	Nex AGI
65	Solar Pro 4Upstage	28	Upstage
66	Claude Sonnet 5 (medium)Anthropic	28	Anthropic
67	Inkling SmallThinking Machines	28	Thinking Machines
68	Qwen3.8-27B (medium)阿里巴巴	28	阿里巴巴
69	GPT-5.6 Terra (low)OpenAI	27	OpenAI
70	JT-4.1 Flash 236B A21BChina Mobile	27	China Mobile
71	Agnes 2.5 Pro AlphaSapiens AI	27	Sapiens AI
72	Quasar 438B (max)Multiverse Computing	27	Multiverse Computing
73	Qwen3.8-27B (low)阿里巴巴	26	阿里巴巴
74	GPT-5.5 InstantOpenAI	26	OpenAI
75	MiMo-V2.5-ProXiaomi	26	Xiaomi
76	Kimi K2.7 CodeMoonshot AI	26	Moonshot AI
77	K2 Horizon MoVA 36B A4BInstitute of Foundation Models	25	Institute of Foundation Models
78	Hy3腾讯AI实验室	25	腾讯AI实验室
79	MiMo-V2.5Xiaomi	25	Xiaomi
80	Qwen3.7-Plus阿里巴巴	25	阿里巴巴
81	GPT-5.6 Luna (medium)OpenAI	25	OpenAI
82	InklingThinking Machines Lab	25	Thinking Machines Lab
83	Ling 3.0 FlashInclusionAI	25	InclusionAI
84	Solar Open2 250BUpstage	25	Upstage
85	Ling-3.0-flash-VLInclusionAI	25	InclusionAI
86	Claude Sonnet 5 (low)Anthropic	24	Anthropic
87	Claude Sonnet 5 (non-reasoning)Anthropic	23	Anthropic
88	Nemotron 3 UltraNVIDIA	23	NVIDIA
89	Qwen2-57B-A14B阿里巴巴	23	阿里巴巴
90	Ling-3.0-flash-FinInclusionAI	23	InclusionAI
91	Gemini 2.0 Flash ExperimentalGoogle DeepMind	22	Google DeepMind
92	Qwen3.8-27B阿里巴巴	22	阿里巴巴
93	Gemini 3.5 Flash-LiteGoogle DeepMind	22	Google DeepMind
94	G9v3-39A5BAI9Stars	22	AI9Stars
95	KAT-Coder-Pro V2KwaiKAT	22	KwaiKAT
96	Qwen3.5 397B A17B (Non-reasoning)Alibaba	21	Alibaba
97	GPT-5.6 Luna (low)OpenAI	21	OpenAI
98	GPT-5.6 Terra (non-reasoning)OpenAI	21	OpenAI
99	K2 Horizon 7BInstitute of Foundation Models	21	Institute of Foundation Models
100	Qwen3.5-Omni-Plus阿里巴巴	20	阿里巴巴
101	OpenAI o3OpenAI	20	OpenAI
102	K-EXAONE 2.0LG AI Research	20	LG AI Research
103	Step 3.7 FlashStepFunAI	19	StepFunAI
104	LongCat 2.0LongCat	19	LongCat
105	Gemma 4 31BGoogle DeepMind	19	Google DeepMind
106	JT-35B-FlashChina Mobile	19	China Mobile
107	Qwen3.5-397B-A17B阿里巴巴	18	阿里巴巴
108	MiMo-V2.5-Pro (Non-reasoning)Xiaomi	18	Xiaomi
109	Qwen3.6-35B-A3B阿里巴巴	18	阿里巴巴
110	Qwen3.5 122B A10B (Non-reasoning)Alibaba	18	Alibaba
111	Muse Glimmer-30B (high)Facebook AI研究实验室	17	Facebook AI研究实验室
112	Doubao Seed CodeByteDance Seed	17	ByteDance Seed
113	Haiku 4.5Anthropic	17	Anthropic
114	Gemma 4 26B A4BGoogle DeepMind	17	Google DeepMind
115	Ring-2.6-1TInclusionAI	17	InclusionAI
116	K2 Horizon 3.7BInstitute of Foundation Models	16	Institute of Foundation Models
117	Qwen3.5-122B-A10B阿里巴巴	16	阿里巴巴
118	GPT-5.6 Luna (non-reasoning)OpenAI	16	OpenAI
119	Haiku 4.5 (non-reasoning)Anthropic	15	Anthropic
120	Ling 3.0 TinyInclusionAI	15	InclusionAI
121	Qwen3.6 35B A3B (Non-reasoning)Alibaba	15	Alibaba
122	Granite 4.2 30BIBM	15	IBM
123	ERNIE 5.0 Thinking Preview百度	14	百度
124	Mistral Medium 3.5MistralAI	14	MistralAI
125	Gemma 4 12BGoogle	14	Google
126	Nova 2 Pro（Preview） (medium)亚马逊	14	亚马逊
127	Gemma 4 31B (Non-reasoning)Google	14	Google
128	Mercury 2Inception	14	Inception
129	Qwen3.5-9B阿里巴巴	14	阿里巴巴
130	Nova 2 Omni（Preview） (medium)亚马逊	14	亚马逊
131	Apriel-v1.6-15B-ThinkerServiceNow	13	ServiceNow
132	Nova 2 Lite (high)亚马逊	13	亚马逊
133	Qwen3.5 9B (Non-reasoning)Alibaba	13	Alibaba
134	EXAONE 4.5 33BLG AI Research	13	LG AI Research
135	Command A+Cohere	13	Cohere
136	Gemma 4 26B A4B (Non-reasoning)Google	13	Google
137	Qwen3.5 4BAlibaba	13	Alibaba
138	Nemotron 3.5 LightningNVIDIA	13	NVIDIA
139	Nemotron 3 SuperNVIDIA	13	NVIDIA
140	Nova 2 Pro（Preview） (low)亚马逊	13	亚马逊
141	Nova 2 Lite (medium)亚马逊	12	亚马逊
142	Qwen3.5-Omni-Flash阿里巴巴	12	阿里巴巴
143	MiniCPM5-2BOpenBMB	12	OpenBMB
144	JT-MINIChina Mobile	12	China Mobile
145	Magistral Medium 1.2Mistral	12	Mistral
146	Nova 2 Lite (low)亚马逊	12	亚马逊
147	HyperNova 60B 2605 (high)Multiverse Computing	12	Multiverse Computing
148	Nemotron Cascade 2 30B A3BNVIDIA	12	NVIDIA
149	GPT OSS 120B (high)OpenAI	12	OpenAI
150	K2 Think V2MBZUAI	11	MBZUAI
151	LongCat Flash LiteLongCat	11	LongCat
152	HyperCLOVA X SEED Think (32B)Naver	11	Naver
153	Mistral Small 4Mistral	11	Mistral
154	Qwen3 Next 80B A3B (Reasoning)Alibaba	11	Alibaba
155	Nova 2 Omni（Preview） (low)亚马逊	11	亚马逊
156	Granite 4.2 8BIBM	11	IBM
157	Mi:dm K 2.5 ProKorea Telecom	11	Korea Telecom
158	G9v3-3BAI9Stars	11	AI9Stars
159	Trinity Large ThinkingArcee AI	11	Arcee AI
160	Qwen3.5 4B (Non-reasoning)Alibaba	11	Alibaba
161	INTELLECT-3Prime Intellect	11	Prime Intellect
162	Solar Open 100BUpstage	10	Upstage
163	Qwen3-Omni-30B-A3B阿里巴巴	10	阿里巴巴
164	GPT OSS 120B (low)OpenAI	10	OpenAI
165	Llama 4 MaverickFacebook AI研究实验室	10	Facebook AI研究实验室
166	Nova 2.0 Pro Preview (Non-reasoning)Amazon	10	Amazon
167	GPT OSS 20B (low)OpenAI	10	OpenAI
168	North Mini CodeCohere	10	Cohere
169	K2-V2 (high)MBZUAI	10	MBZUAI
170	Qwen3-Next阿里巴巴	10	阿里巴巴
171	DiffusionGemma 26B A4BGoogle	10	Google
172	Gemma 4 12B (Non-reasoning)Google	9	Google
173	Mistral Large 3MistralAI	9	MistralAI
174	Qwen3-Coder-Next阿里巴巴	9	阿里巴巴
175	Motif-2-12.7BMotif Technologies	9	Motif Technologies
176	Nova PremierAmazon	9	Amazon
177	Granite 4.2 3BIBM	9	IBM
178	K2-V2 (medium)MBZUAI	9	MBZUAI
179	Llama Nemotron Super 49B v1.5Meta	9	Meta
180	Mistral Small 4 (Non-reasoning)Mistral	9	Mistral
181	Tri-21B-ThinkTrillion Labs	9	Trillion Labs
182	GPT OSS 20B (high)OpenAI	9	OpenAI
183	Gemma 4 E4BGoogle DeepMind	9	Google DeepMind
184	Nemotron 3 NanoNVIDIA	9	NVIDIA
185	MiniCPM5-1BOpenBMB	9	OpenBMB
186	Sarvam 105B (high)Sarvam	9	Sarvam
187	Nova 2.0 Lite (Non-reasoning)Amazon	9	Amazon
188	MiniCPM5-1B (Non-reasoning)OpenBMB	9	OpenBMB
189	Devstral 2Mistral	9	Mistral
190	Magistral Small 1.2Mistral	9	Mistral
191	Nanbeige4.1-3BNanbeige	8	Nanbeige
192	LFM2.5-2.6BLiquid AI	8	Liquid AI
193	EXAONE 4.0 32BLG AI Research	8	LG AI Research
194	Nova 2.0 Omni (Non-reasoning)Amazon	8	Amazon
195	Llama 4 ScoutFacebook AI研究实验室	8	Facebook AI研究实验室
196	Hermes 4 70BNous Research	8	Nous Research
197	Falcon-H1R-7BTII UAE	8	TII UAE
198	Gemma 4 E2BGoogle DeepMind	8	Google DeepMind
199	Qwen3 Omni 30B A3B (Reasoning)Alibaba	8	Alibaba
200	Step3 VL 10BStepFun	8	StepFun
201	Llama3.3-70B-InstructFacebook AI研究实验室	8	Facebook AI研究实验室
202	Devstral Small 2Mistral	8	Mistral
203	Llama Nemotron UltraNVIDIA	8	NVIDIA
204	ERNIE-4.5-300B-A47B百度	8	百度
205	Hermes 4 405BNous Research	7	Nous Research
206	NVIDIA Nemotron Nano 12B v2 VLNVIDIA	7	NVIDIA
207	Gemma 4 E4B (Non-reasoning)Google	7	Google
208	NVIDIA Nemotron Nano 9B V2NVIDIA	7	NVIDIA
209	Hermes 4 405B (Non-reasoning)Nous Research	7	Nous Research
210	Nemotron 3 Nano 4BNVIDIA	7	NVIDIA
211	Llama Nemotron Super 49B v1.5 (Non-reasoning)NVIDIA	7	NVIDIA
212	K2-V2 (low)MBZUAI	7	MBZUAI
213	Kimi Linear 48B A3B InstructKimi	7	Kimi
214	Llama3.1-405BFacebook AI研究实验室	7	Facebook AI研究实验室
215	LFM2.5-8B-A1BLiquid AI	7	Liquid AI
216	Ring-flash-2.0InclusionAI	7	InclusionAI
217	Olmo 3.1 32B ThinkAI2	7	AI2
218	C4AI Command A (202503)CohereAI	7	CohereAI
219	Qwen3.5 2BAlibaba	7	Alibaba
220	Llama 3.1 Nemotron 70BNVIDIA	7	NVIDIA
221	Nemotron 3 Nano (Non-reasoning)NVIDIA	7	NVIDIA
222	NVIDIA Nemotron Nano 9B V2 (Non-reasoning)NVIDIA	7	NVIDIA
223	Hermes 4 70B (Non-reasoning)Nous Research	7	Nous Research
224	Sarvam 30B (high)Sarvam	7	Sarvam
225	Olmo 3.1 32B InstructAI2	6	AI2
226	Gemma 4 E2B (Non-reasoning)Google	6	Google
227	R1 1776Perplexity	6	Perplexity
228	Llama 3.2-Vision-90BFacebook AI研究实验室	6	Facebook AI研究实验室
229	Celeris-1Celeris	6	Celeris
230	Phi-4-mini-instruct (3.8B)Microsoft Azure	6	Microsoft Azure
231	EXAONE 4.0 32B (Non-reasoning)LG AI Research	6	LG AI Research
232	Qwen3.5 2B (Non-reasoning)Alibaba	6	Alibaba
233	Qwen3.5 0.8BAlibaba	6	Alibaba
234	DeepHermes 3 - Mistral 24BNous Research	6	Nous Research
235	Jamba 1.7 LargeAI21 Labs	6	AI21 Labs
236	Granite 4.0 H SmallIBM	6	IBM
237	Ministral 3 14BMistralAI	6	MistralAI
238	Qwen3-Omni-30B-A3B阿里巴巴	6	阿里巴巴
239	LFM2 24B A2BLiquid AI	6	Liquid AI
240	Phi-4-reasoningMicrosoft Azure	6	Microsoft Azure
241	Amazon Nova Micro亚马逊	6	亚马逊
242	NVIDIA Nemotron Nano 12B v2 VL (Non-reasoning)NVIDIA	6	NVIDIA
243	Phi-4-multimodal-instruct Microsoft Azure	6	Microsoft Azure
244	MiniCPM-V 4.6 1.3BOpenBMB	6	OpenBMB
245	Jamba Reasoning 3BAI21 Labs	6	AI21 Labs
246	Reka Flash 3Reka AI	6	Reka AI
247	Olmo 3 7B ThinkAI2	6	AI2
248	Molmo 7B-DAllen Institute for AI	6	Allen Institute for AI
249	Ministral 3 8BMistralAI	5	MistralAI
250	Llama 3.2-Vision-11BFacebook AI研究实验室	5	Facebook AI研究实验室
251	Qwen3.5 0.8B (Non-reasoning)Alibaba	5	Alibaba
252	Exaone 4.0 1.2BLG AI Research	5	LG AI Research
253	Olmo 3 7BAI2	5	AI2
254	Exaone 4.0 1.2B (Non-reasoning)LG AI Research	5	LG AI Research
255	LFM2.5-1.2B-ThinkingLiquid AI	5	Liquid AI
256	Jamba 1.7 MiniAI21 Labs	5	AI21 Labs
257	LFM2.5-1.2B-InstructLiquid AI	5	Liquid AI
258	Granite 4.0 H 1BIBM	5	IBM
259	Gemma 3-270MGoogle DeepMind	5	Google DeepMind
260	Apertus 70B InstructSwiss AI	5	Swiss AI
261	Granite 4.0 MicroIBM	5	IBM
262	DeepHermes 3 - Llama-3.1 8BNous Research	5	Nous Research
263	Molmo2-8BAI2	5	AI2
264	Ministral 3 3BMistral	5	Mistral
265	LFM2.5-VL-1.6BLiquid AI	5	Liquid AI
266	Granite 4.0 350MIBM	5	IBM
267	Tiny Aya GlobalCohere	5	Cohere
268	Apertus 8B InstructSwiss AI	5	Swiss AI
269	Granite 4.0 H 350MIBM	5	IBM
270	K2 Horizon 0.9BInstitute of Foundation Models	3	Institute of Foundation Models

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。