模型排行

排名	變化	模型	得分	置信區間	投票	價格 $/M
1	—	kimi-k3-max開源Moonshot AI	1,488	±5	28,280	—
2	—	mimo-v2.6-pro開源Xiaomi	1,480	±9	4,056	$0.54
3	—	glm-5.3-max開源Z.ai	1,478	±6	17,857	—
4	▲ 1	glm-5.2-max開源Z.ai	1,476	±4	45,399	—
5	▼ 1	deepseek-v4.1-flash-max開源DeepSeek	1,474	±7	8,728	—
6	—	glm-5.3-flash開源Z.ai	1,473	±5	22,971	$0.24
7	—	mimo-v2.5-pro開源Xiaomi	1,468	±4	70,871	—
8	—	glm-5.1開源Z.ai	1,465	±4	58,358	—
9	—	deepseek-v4-pro-high-20260813開源DeepSeek · 另有 1 個檔位	1,464	±7	9,780	—
10	—	kimi-k2.6開源Moonshot AI	1,461	±4	39,936	—
11	—	deepseek-v4-pro開源DeepSeek	1,458	±4	57,478	—
12	—	glm-5開源Z.ai	1,458	±4	29,174	—
13	—	hy3開源Tencent	1,456	±7	10,476	—
14	▲ 1	gemma-4-31b開源Google	1,453	±7	6,133	$0.000
15	▼ 1	mimo-v2.6-flash開源Xiaomi	1,452	±8	6,053	$0.18
16	—	kimi-k2.5-thinking開源Moonshot AI	1,450	±3	74,565	—
17	▲ 1	qwen3.5-397b-a17b開源Alibaba	1,442	±3	88,660	—
18	▲ 1	glm-4.7開源Z.ai	1,441	±6	12,228	—
19	▼ 2	inkling開源Thinking Machines	1,441	±5	35,204	$1.8
20	—	minimax-m3開源MiniMax	1,440	±4	58,757	$0.52
21	—	deepseek-v4-flash-high-preview開源DeepSeek	1,438	±4	51,648	—
22	—	qwen3.8-27b開源Alibaba	1,438	±5	18,462	$1.1
23	—	gemma-4-26b-a4b開源Google	1,437	±7	6,081	—
24	—	deepseek-v4-flash開源DeepSeek	1,436	±4	51,850	—
25	—	mimo-v2.5開源Xiaomi	1,434	±4	47,266	—
26	—	kimi-k2.5-instant開源Moonshot AI	1,431	±7	8,356	—
27	—	kimi-k2-thinking-turbo開源Moonshot AI	1,430	±3	63,210	—
28	—	mistral-medium-3.5開源Mistral	1,427	±6	11,757	$3.0
29	—	nvidia-nemotron-3-ultra-550b-a55b-nvfp4開源NVIDIA	1,426	±7	11,515	—
30	▲ 2	muse-glimmer開源Meta	1,425	±10	3,893	$0.52
31	▼ 1	deepseek-v3.2-exp-thinking開源DeepSeek · 另有 1 個檔位	1,425	±7	8,947	—
32	▼ 1	deepseek-v3.2開源DeepSeek · 另有 1 個檔位	1,425	±3	47,999	—
33	—	glm-4.6開源Z.ai	1,424	±4	35,662	—
34	—	qwen3-235b-a22b-instruct-2507開源Alibaba	1,422	±3	97,788	—
35	—	deepseek-r1-0528開源DeepSeek · 另有 1 個檔位	1,422	±6	18,059	—
36	—	kimi-k2-0905-preview開源Moonshot AI · 另有 1 個檔位	1,419	±7	11,531	—
37	—	deepseek-v3.1-terminus-thinking開源DeepSeek · 另有 1 個檔位	1,417	±10	3,354	—
38	—	deepseek-v3.1開源DeepSeek · 另有 1 個檔位	1,417	±6	14,540	—
39	—	qwen3.5-122b-a10b開源Alibaba	1,416	±4	29,651	—
40	—	minimax-m2.7開源MiniMax	1,415	±3	79,944	—
41	▲ 1	mistral-large-3開源Mistral	1,414	±3	79,452	—
42	▼ 1	qwen3-vl-235b-a22b-instruct開源Alibaba	1,413	±7	11,241	—
43	—	glm-4.5開源Z.ai	1,411	±5	23,661	—
44	—	hunyuan-hy3-preview開源Tencent	1,409	±7	6,874	—
45	—	qwen3.5-27b開源Alibaba	1,409	±4	28,612	—
46	—	Inkling Small開源Thinking Machines	1,405	±5	27,070	$0.52
47	—	qwen3-235b-a22b-no-thinking開源Alibaba	1,403	±5	37,279	—
48	—	longcat-flash-chat開源Meituan	1,401	±6	11,123	—
49	—	qwen3-235b-a22b-thinking-2507開源Alibaba	1,400	±7	8,774	—
50	—	qwen3-next-80b-a3b-instruct開源Alibaba	1,399	±5	22,603	—
51	—	deepseek-v3-0324開源DeepSeek · 另有 1 個檔位	1,396	±4	44,758	—
52	—	qwen3-vl-235b-a22b-thinking開源Alibaba	1,395	±7	7,805	—
53	—	qwen3.5-35b-a3b開源Alibaba	1,394	±4	30,408	—
54	—	step-3.5-flash開源StepFun	1,393	±4	60,207	—
55	—	mimo-v2-flash (non-thinking)開源Xiaomi · 另有 1 個檔位	1,392	±3	47,998	—
56	—	minimax-m2.5開源MiniMax	1,391	±4	42,730	—
57	—	qwen3-coder-480b-a35b-instruct開源Alibaba	1,387	±5	25,006	—
58	—	minimax-m2.1-preview開源MiniMax	1,385	±5	17,187	—
59	—	qwen3-30b-a3b-instruct-2507開源Alibaba	1,383	±5	23,174	—
60	▲ 1	glm-4.6v開源Z.ai	1,379	±11	2,830	—
61	▼ 1	trinity-large-preview開源 · 另有 1 個檔位	1,379	±4	31,201	—
62	—	qwen3-235b-a22b開源Alibaba	1,375	±5	25,825	—
63	—	glm-4.5-air開源Z.ai	1,373	±4	30,454	—
64	—	qwen3-next-80b-a3b-thinking開源Alibaba	1,369	±6	13,371	—
65	—	gemma-3-27b-it開源Google	1,365	±4	46,425	—
66	—	glm-4.7-flash開源Z.ai	1,365	±6	11,977	—
67	—	minimax-m1開源MiniMax	1,364	±4	34,042	—
68	—	nvidia-nemotron-3-super-120b-a12b開源NVIDIA	1,360	±7	7,859	—
69	—	mistral-small-2506開源Mistral	1,356	±5	17,348	—
70	—	intellect-3開源	1,356	±8	5,417	—
71	—	command-a-03-2025開源Cohere	1,354	±3	55,467	—
72	—	glm-4.5v開源Z.ai	1,352	±8	4,807	—
73	—	gpt-oss-120b開源OpenAI	1,352	±4	30,018	$0.26
74	▲ 3	nvidia-nemotron-3.5-lightning-30b-a3b-nvfp4開源NVIDIA	1,350	±6	14,877	—
75	▼ 1	step-3開源StepFun	1,349	±8	6,266	—
76	▼ 1	llama-3.1-nemotron-ultra-253b-v1開源NVIDIA	1,348	±12	2,549	—
77	▼ 1	qwen3-32b開源Alibaba	1,347	±9	3,926	—
78	—	ling-flash-2.0開源Ant Group	1,344	±7	6,887	—
79	—	minimax-m2開源MiniMax	1,343	±8	6,817	—
80	—	nvidia-llama-3.3-nemotron-super-49b-v1.5開源NVIDIA	1,343	±10	3,308	—
81	—	gemma-3-12b-it開源Google	1,342	±10	3,829	—
82	—	granite-4.2-30b開源IBM	1,340	±10	3,401	—
83	—	qwq-32b開源Alibaba · 另有 1 個檔位	1,336	±4	25,078	—
84	—	llama-3.1-405b-instruct-bf16開源Meta	1,335	±4	41,375	—
85	—	llama-3.1-405b-instruct-fp8開源Meta	1,334	±4	59,656	—
86	—	olmo-3.1-32b-instruct開源Ai2	1,329	±6	11,801	—
87	—	llama-3.3-nemotron-49b-super-v1開源NVIDIA	1,328	±12	2,218	—
88	—	llama-4-maverick-17b-128e-instruct開源Meta	1,327	±4	39,262	—
89	—	qwen3-30b-a3b開源Alibaba	1,326	±5	26,037	—
90	—	deepseek-v2.5-1210開源DeepSeek · 另有 1 個檔位	1,323	±8	6,795	—
91	—	ring-flash-2.0開源Ant Group	1,323	±7	6,932	—
92	—	molmo-2-8b開源Ai2	1,322	±21	818	—
93	—	llama-4-scout-17b-16e-instruct開源Meta	1,322	±5	29,667	—
94	—	qwen-max-0919開源Alibaba	1,318	±6	16,478	—
95	—	llama-3.3-70b-instruct開源Meta	1,318	±4	54,366	—
96	—	gpt-oss-20b開源OpenAI	1,318	±6	10,368	—
97	—	gemma-3n-e4b-it開源Google	1,317	±5	22,042	—
98	—	mistral-large-2407開源Mistral · 另有 1 個檔位	1,314	±4	45,459	—
99	—	athene-v2-chat開源	1,314	±5	24,739	—
100	—	nvidia-nemotron-3-nano-30b-a3b-bf16開源NVIDIA	1,314	±5	15,722	—
101	—	athene-70b-0725開源	1,307	±6	19,621	—
102	—	olmo-3-32b-think開源Ai2	1,306	±8	5,811	—
103	—	granite-4.1-8b開源IBM	1,304	±10	4,161	—
104	—	gemma-3-4b-it開源Google	1,303	±9	4,171	—
105	—	mistral-small-3.1-24b-instruct-2503開源Mistral	1,303	±5	32,551	—
106	—	qwen2.5-72b-instruct開源Alibaba	1,303	±4	39,406	—
107	—	llama-3.1-nemotron-70b-instruct開源NVIDIA	1,299	±8	7,140	—
108	—	llama-3.1-70b-instruct開源Meta	1,293	±4	55,240	—
109	—	granite-4.2-3b開源IBM	1,290	±12	3,247	—
110	—	gemma-2-27b-it開源Google	1,289	±3	75,754	—
111	—	jamba-1.5-large開源	1,289	±7	8,662	—
112	—	granite-4.2-8b開源IBM	1,288	±11	3,237	$0.11
113	—	ibm-granite-h-small開源IBM	1,287	±8	5,576	—
114	—	llama-3.1-nemotron-51b-instruct開源NVIDIA	1,287	±10	3,749	—
115	—	olmo-3.1-32b-think開源Ai2	1,286	±7	8,278	—
116	—	llama-3.1-tulu-3-70b開源Ai2	1,286	±10	2,846	—
117	—	gemma-2-9b-it-simpo開源	1,280	±7	10,072	—
118	—	nemotron-4-340b-instruct開源NVIDIA	1,277	±5	19,659	—
119	—	llama-3-70b-instruct開源Meta	1,277	±4	156,876	—
120	—	command-r-plus-08-2024開源Cohere	1,276	±7	9,866	—
121	—	mistral-small-24b-instruct-2501開源Mistral	1,274	±6	14,681	—
122	—	qwen2.5-coder-32b-instruct開源Alibaba	1,271	±8	5,432	—
123	—	c4ai-aya-expanse-32b開源Cohere	1,267	±5	27,124	—
124	—	gemma-2-9b-it開源Google	1,267	±4	54,611	—
125	—	deepseek-coder-v2開源DeepSeek	1,265	±6	15,147	—
126	—	qwen2-72b-instruct開源Alibaba	1,262	±5	37,325	—
127	—	command-r-plus開源Cohere	1,262	±4	77,554	—
128	—	phi-4開源Microsoft	1,256	±5	24,126	—
129	—	olmo-2-0325-32b-instruct開源Ai2	1,252	±11	3,334	—
130	—	command-r-08-2024開源Cohere	1,250	±7	10,140	—
131	—	jamba-1.5-mini開源	1,240	±7	8,858	—
132	—	ministral-8b-2410開源Mistral	1,238	±9	4,781	—
133	—	qwen1.5-110b-chat開源Alibaba	1,234	±6	26,195	—
134	—	qwen1.5-72b-chat開源Alibaba	1,233	±5	39,302	—
135	—	mixtral-8x22b-instruct-v0.1開源Mistral	1,230	±5	51,416	—
136	—	command-r開源Cohere	1,227	±5	54,036	—
137	—	llama-3-8b-instruct開源Meta	1,224	±4	104,642	—
138	—	c4ai-aya-expanse-8b開源Cohere	1,223	±7	9,818	—
139	—	llama-3.1-tulu-3-8b開源Ai2	1,220	±11	2,896	—
140	—	zephyr-orpo-141b-A35b-v0.1開源	1,213	±11	4,652	—
141	—	yi-1.5-34b-chat開源	1,213	±5	24,146	—
142	—	llama-3.1-8b-instruct開源Meta	1,211	±4	49,605	—
143	—	granite-3.1-8b-instruct開源IBM	1,208	±11	3,090	—
144	—	qwen1.5-32b-chat開源Alibaba	1,204	±6	21,741	—
145	—	gemma-2-2b-it開源Google	1,200	±4	46,616	—
146	—	phi-3-medium-4k-instruct開源Microsoft	1,198	±5	25,055	—
147	—	mixtral-8x7b-instruct-v0.1開源Mistral	1,197	±4	73,503	—
148	—	dbrx-instruct-preview開源	1,195	±6	32,191	—
149	—	qwen1.5-14b-chat開源Alibaba	1,191	±7	17,839	—
150	—	internlm2_5-20b-chat開源	1,191	±7	9,901	—
151	—	deepseek-llm-67b-chat開源DeepSeek	1,185	±11	4,932	—
152	—	wizardlm-70b開源Microsoft	1,185	±9	8,214	—
153	—	yi-34b-chat開源	1,184	±7	15,483	—
154	—	granite-3.0-8b-instruct開源IBM	1,183	±9	6,638	—
155	—	openchat-3.5開源 · 另有 1 個檔位	1,183	±10	7,968	—
156	—	gemma-1.1-7b-it開源Google	1,183	±6	23,893	—
157	—	snowflake-arctic-instruct開源	1,180	±6	32,832	—
158	—	granite-3.1-2b-instruct開源IBM	1,179	±11	3,188	—
159	—	tulu-2-dpo-70b開源	1,178	±10	6,535	—
160	—	openhermes-2.5-mistral-7b開源	1,176	±10	5,006	—
161	—	vicuna-33b開源	1,173	±6	22,479	—
162	—	phi-3-small-8k-instruct開源Microsoft	1,171	±6	17,766	—
163	—	starling-lm-7b-beta開源	1,171	±7	16,056	—
164	—	llama-2-70b-chat開源Meta	1,171	±5	38,492	—
165	—	starling-lm-7b-alpha開源	1,167	±8	10,224	—
166	—	llama-3.2-3b-instruct開源Meta	1,167	±8	7,936	—
167	—	nous-hermes-2-mixtral-8x7b-dpo開源	1,164	±12	3,777	—
168	—	granite-3.0-2b-instruct開源IBM	1,157	±8	6,837	—
169	—	llama2-70b-steerlm-chat開源NVIDIA	1,155	±13	3,585	—
170	—	solar-10.7b-instruct-v1.0開源	1,152	±13	4,155	—
171	—	dolphin-2.2.1-mistral-7b開源	1,152	±15	1,679	—
172	—	mpt-30b-chat開源	1,151	±12	2,572	—
173	—	wizardlm-13b開源Microsoft	1,149	±9	7,044	—
174	—	mistral-7b-instruct-v0.2開源Mistral	1,149	±7	19,402	—
175	—	falcon-180b-chat開源	1,148	±17	1,295	—
176	—	qwen1.5-7b-chat開源Alibaba	1,144	±10	4,737	—
177	—	phi-3-mini-4k-instruct-june-2024開源Microsoft	1,143	±6	12,297	—
178	—	vicuna-13b開源	1,142	±7	19,367	—
179	—	llama-2-13b-chat開源Meta	1,141	±7	19,174	—
180	—	qwen-14b-chat開源Alibaba	1,139	±11	4,964	—
181	—	gemma-7b-it開源Google	1,138	±9	8,925	—
182	—	codellama-34b-instruct開源Meta	1,137	±9	7,366	—
183	—	zephyr-7b-beta開源	1,131	±9	11,118	—
184	—	phi-3-mini-128k-instruct開源Microsoft	1,130	±7	20,685	—
185	—	phi-3-mini-4k-instruct開源Microsoft	1,128	±6	20,118	—
186	—	guanaco-33b開源	1,127	±12	2,921	—
187	—	zephyr-7b-alpha開源	1,127	±16	1,785	—
188	—	stripedhyena-nous-7b開源	1,122	±11	5,182	—
189	—	codellama-70b-instruct開源Meta	1,119	±18	1,143	—
190	—	gemma-1.1-2b-it開源Google	1,117	±8	10,854	—
191	—	vicuna-7b開源	1,115	±9	6,923	—
192	—	smollm2-1.7b-instruct開源	1,115	±14	2,199	—
193	—	llama-3.2-1b-instruct開源Meta	1,111	±8	8,045	—
194	—	mistral-7b-instruct開源Mistral	1,110	±9	8,977	—
195	—	llama-2-7b-chat開源Meta	1,108	±7	14,148	—
196	—	gemma-2b-it開源Google	1,094	±11	4,780	—
197	—	qwen1.5-4b-chat開源Alibaba	1,091	±9	7,597	—
198	—	olmo-7b-instruct開源Ai2	1,074	±11	6,328	—
199	—	koala-13b開源	1,071	±10	6,965	—
200	—	alpaca-13b開源	1,070	±11	5,745	—
201	—	gpt4all-13b-snoozy開源	1,068	±15	1,743	—
202	—	mpt-7b-chat開源	1,063	±12	3,924	—
203	—	chatglm3-6b開源	1,057	±12	4,658	—
204	—	RWKV-4-Raven-14B開源	1,042	±11	4,845	—
205	—	chatglm2-6b開源	1,025	±14	2,658	—
206	—	oasst-pythia-12b開源	1,023	±11	6,310	—
207	—	chatglm-6b開源	996	±12	4,914	—
208	—	fastchat-t5-3b開源	992	±12	4,203	—
209	—	dolly-v2-12b開源	982	±13	3,412	—
210	—	llama-13b開源Meta	975	±16	2,391	—
211	—	stablelm-tuned-alpha-7b開源	953	±13	3,287	—

其餘 161 個模型

智力進步

每家廠商當時最強的模型在 Artificial Analysis 智力指數上的分數，按模型釋出的月份畫。

漲得最多 · Anthropic +57.2目前最高 · Anthropic 64.6最高與最低相差 18.1 分時間範圍 2025年11月 – 2026年11月

64.6Anthropic60.0Google56.9OpenAI50.1Meta46.5Kimi