模型排行

排名	變化	模型	得分	置信區間	投票	價格 $/M
1	—	llama-3.1-405b-instruct-fp8開源Meta	1,252	±3	45,768	—
2	—	llama-3.1-405b-instruct-bf16開源Meta	1,251	±5	14,608	—
3	—	mistral-large-2407開源Mistral	1,230	±3	42,110	—
4	—	qwen-max-0919開源Alibaba	1,229	±4	17,658	—
5	—	deepseek-v2.5開源DeepSeek	1,223	±4	20,034	—
6	▲ 1	qwen2.5-72b-instruct開源Alibaba	1,221	±4	20,113	—
7	▼ 1	athene-70b-0725開源NexusFlow	1,221	±5	20,709	—
8	—	llama-3.1-nemotron-70b-instruct開源NVIDIA	1,215	±6	6,572	—
9	▲ 1	llama-3.1-70b-instruct開源Meta	1,211	±3	41,484	—
10	▲ 1	gemma-2-27b-it開源Google	1,205	±3	52,532	—
11	▲ 1	jamba-1.5-large開源AI21 Labs	1,203	±6	9,188	—
12	▼ 3	llama-3.1-nemotron-51b-instruct開源NVIDIA	1,201	±9	3,153	—
13	—	gemma-2-9b-it-simpo開源Princeton	1,198	±6	10,618	—
14	—	command-r-plus-08-2024開源Cohere	1,196	±5	10,602	—
15	—	nemotron-4-340b-instruct開源NVIDIA	1,196	±4	20,645	—
16	—	llama-3-70b-instruct開源Meta	1,194	±3	163,936	—
17	—	command-r-plus開源Cohere	1,183	±3	80,907	—
18	—	deepseek-coder-v2開源DeepSeek	1,181	±4	15,798	—
19	—	qwen2-72b-instruct開源Alibaba	1,180	±3	38,985	—
20	—	gemma-2-9b-it開源Google	1,176	±3	37,949	—
21	—	command-r-08-2024開源Cohere	1,175	±5	10,931	—
22	—	jamba-1.5-mini開源AI21 Labs	1,158	±6	9,298	—
23	—	qwen1.5-110b-chat開源Alibaba	1,155	±4	27,474	—
24	—	mixtral-8x22b-instruct-v0.1開源Mistral	1,149	±3	53,841	—
25	—	command-r開源Cohere	1,148	±3	56,405	—
26	—	qwen1.5-72b-chat開源Alibaba	1,148	±3	40,646	—
27	—	llama-3-8b-instruct開源Meta	1,142	±3	109,373	—
28	—	yi-1.5-34b-chat開源01 AI	1,132	±4	25,143	—
29	—	zephyr-orpo-141b-A35b-v0.1開源HuggingFace	1,131	±6	4,857	—
30	—	llama-3.1-8b-instruct開源Meta	1,129	±3	39,806	—
31	—	qwen1.5-32b-chat開源Alibaba	1,126	±5	22,760	—
32	—	phi-3-medium-4k-instruct開源Microsoft	1,118	±3	26,149	—
33	—	internlm2_5-20b-chat開源InternLM	1,116	±6	10,716	—
34	—	mixtral-8x7b-instruct-v0.1開源Mistral	1,114	±0	76,141	—
35	—	dbrx-instruct-preview開源Databricks	1,113	±3	33,725	—
36	—	gemma-2-2b-it開源Google	1,111	±4	29,470	—
37	—	qwen1.5-14b-chat開源Alibaba	1,107	±4	18,669	—
38	—	wizardlm-70b開源Microsoft	1,104	±6	8,382	—
39	—	snowflake-arctic-instruct開源Snowflake	1,103	±4	34,192	—
40	—	yi-34b-chat開源01 AI	1,101	±4	15,928	—
41	—	openchat-3.5-0106開源OpenChat · 另有 1 個檔位	1,100	±6	12,973	—
42	—	phi-3-small-8k-instruct開源Microsoft	1,098	±4	18,497	—
43	—	gemma-1.1-7b-it開源Google	1,098	±4	25,090	—
44	—	llama-3.2-3b-instruct開源Meta	1,093	±7	8,467	—
45	—	tulu-2-dpo-70b開源AllenAI/UW	1,092	±7	6,666	—
46	—	vicuna-33b開源LMSYS	1,091	±4	22,945	—
47	▲ 1	starling-lm-7b-beta開源Nexusflow	1,091	±5	16,670	—
48	▼ 1	deepseek-llm-67b-chat開源DeepSeek	1,091	±8	4,983	—
49	—	openhermes-2.5-mistral-7b開源NousResearch	1,090	±8	5,092	—
50	—	starling-lm-7b-alpha開源UC Berkeley	1,083	±5	10,415	—
51	—	llama-2-70b-chat開源Meta	1,081	±3	39,621	—
52	—	llama2-70b-steerlm-chat開源NVIDIA	1,074	±10	3,638	—
53	—	nous-hermes-2-mixtral-8x7b-dpo開源NousResearch	1,071	±8	3,836	—
54	—	dolphin-2.2.1-mistral-7b開源Cognitive Computations	1,069	±13	1,713	—
55	—	phi-3-mini-4k-instruct-june-2024開源Microsoft	1,068	±6	12,881	—
56	—	mistral-7b-instruct-v0.2開源Mistral	1,067	±5	20,065	—
57	—	solar-10.7b-instruct-v1.0開源Upstage AI	1,066	±9	4,287	—
58	—	qwen1.5-7b-chat開源Alibaba	1,065	±7	4,865	—
59	—	wizardlm-13b開源Microsoft	1,063	±8	7,185	—
60	—	mpt-30b-chat開源MosaicML	1,060	±9	2,649	—
61	—	falcon-180b-chat開源TII	1,059	±17	1,326	—
62	—	vicuna-13b開源LMSYS	1,059	±4	19,788	—
63	—	qwen-14b-chat開源Alibaba	1,057	±7	5,068	—
64	—	phi-3-mini-4k-instruct開源Microsoft	1,055	±4	21,119	—
65	—	llama-2-13b-chat開源Meta	1,053	±4	19,727	—
66	—	zephyr-7b-beta開源HuggingFace	1,049	±6	11,327	—
67	—	phi-3-mini-128k-instruct開源Microsoft	1,049	±5	21,621	—
68	▲ 1	codellama-34b-instruct開源Meta	1,047	±7	7,508	—
69	▼ 1	gemma-7b-it開源Google	1,047	±5	9,175	—
70	—	zephyr-7b-alpha開源HuggingFace	1,046	±14	1,813	—
71	—	llama-3.2-1b-instruct開源Meta	1,044	±7	8,578	—
72	—	guanaco-33b開源UW	1,040	±11	2,998	—
73	—	gemma-1.1-2b-it開源Google	1,039	±6	11,347	—
74	—	codellama-70b-instruct開源Meta	1,038	±18	1,189	—
75	—	stripedhyena-nous-7b開源Together AI	1,034	±9	5,271	—
76	—	mistral-7b-instruct開源Mistral	1,027	±5	9,143	—
77	—	llama-2-7b-chat開源Meta	1,024	±5	14,556	—
78	—	vicuna-7b開源LMSYS	1,022	±6	7,017	—
79	—	qwen1.5-4b-chat開源Alibaba	1,015	±6	7,812	—
80	—	gemma-2b-it開源Google	1,008	±9	4,919	—
81	—	olmo-7b-instruct開源Ai2	993	±7	6,499	—
82	—	koala-13b開源UC Berkeley	980	±7	7,033	—
83	—	chatglm3-6b開源Tsinghua	975	±8	4,762	—
84	—	gpt4all-13b-snoozy開源Nomic AI	972	±12	1,786	—
85	—	alpaca-13b開源Stanford	968	±8	5,872	—
86	—	mpt-7b-chat開源MosaicML	967	±10	4,013	—
87	—	RWKV-4-Raven-14B開源RWKV	949	±9	4,934	—
88	—	chatglm2-6b開源Tsinghua	942	±12	2,707	—
89	—	oasst-pythia-12b開源OpenAssistant	932	±9	6,382	—
90	—	fastchat-t5-3b開源LMSYS	896	±9	4,302	—
91	—	chatglm-6b開源Tsinghua	888	±11	4,995	—
92	—	dolly-v2-12b開源Databricks	882	±11	3,486	—
93	—	llama-13b開源Meta	868	±14	2,445	—
94	—	stablelm-tuned-alpha-7b開源Stability AI	859	±12	3,336	—

其餘 44 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