모델 순위

순위	변동	모델	지능	가격 $/M	속도 tok/s	첫 토큰
1	—	Inkling (xhigh)오픈Thinking Machines	42.3	$1.8	84	1.7초
2	—	Inkling Small오픈Thinking Machines	41.2	$0.52	64	1.8초
3	—	Nemotron 3 Ultra 550B A55B (Reasoning)오픈NVIDIA	38.3	$1.1	140	1.8초
4	신규	Muse Glimmer (high)오픈Meta	35.1	$0.58	111	0.9초
5	▼ 1	Gemma 4 31B (Reasoning)오픈Google · 변형 1개 더	29.7	$0.000	36	1.1초
6	▼ 1	Gemma 4 26B A4B (Reasoning)오픈Google · 변형 1개 더	26.1	$0.20	—	—
7	▼ 1	Nemotron 3 Super 120B A12B (Reasoning)오픈NVIDIA	25.7	$0.35	132	1.9초
8	▼ 1	gpt-oss-120b (high)오픈OpenAI · 변형 1개 더	24.1	$0.26	177	0.8초
9	신규	Nemotron 3.5 Lightning오픈NVIDIA	23.6	$0.088	298	1.0초
10	▼ 2	Gemma 4 12B (Reasoning)오픈Google · 변형 1개 더	22.2	$0.15	110	2.3초
11	▼ 2	Apriel-v1.5-15B-Thinker오픈ServiceNow	21.6	$0.000	—	—
12	▼ 2	Apriel-v1.6-15B-Thinker오픈ServiceNow	20.8	$0.000	—	—
13	▼ 2	Nemotron Cascade 2 30B A3B오픈NVIDIA	17.8	—	—	—
14	▼ 2	gpt-oss-20b (high)오픈OpenAI · 변형 1개 더	15.2	$0.093	125	1.0초
15	▼ 2	Nemotron 3 Nano Omni 30B A3B Reasoning오픈NVIDIA	15.0	$0.13	324	1.1초
16	▼ 2	NVIDIA Nemotron 3 Nano 30B A3B (Reasoning)오픈NVIDIA · 변형 1개 더	14.5	$0.088	219	1.3초
17	▼ 2	Llama 4 Maverick오픈Meta	14.5	$0.42	121	1.0초
18	▼ 2	DiffusionGemma 26B A4B오픈Google	13.5	—	—	—
19	▼ 2	Llama Nemotron Super 49B v1.5 (Reasoning)오픈NVIDIA · 변형 1개 더	12.4	$0.40	115	3.8초
20	▼ 1	Llama 3.3 Nemotron Super 49B v1 (Reasoning)오픈NVIDIA · 변형 1개 더	12.2	—	—	—
21	▼ 3	Gemma 4 E4B (Reasoning)오픈Google · 변형 1개 더	12.2	$0.040	50	1.0초
22	▼ 2	Llama 4 Scout오픈Meta	10.3	$0.30	136	0.8초
23	▼ 2	Gemma 4 E2B (Reasoning)오픈Google · 변형 1개 더	9.5	—	—	—
24	▼ 2	Llama 3.3 Instruct 70B오픈Meta	9.3	$0.67	90	1.6초
25	▼ 1	Llama 3.1 Nemotron Ultra 253B v1 (Reasoning)오픈NVIDIA	8.9	$0.90	53	2.5초
26	—	NVIDIA Nemotron Nano 12B v2 VL (Reasoning)오픈NVIDIA · 변형 1개 더	8.8	$0.30	138	3.7초
27	▼ 4	Granite 4.1 30B오픈IBM	8.7	—	—	—
28	▼ 1	NVIDIA Nemotron Nano 9B V2 (Reasoning)오픈NVIDIA · 변형 1개 더	8.7	$0.070	175	3.8초
29	▼ 4	NVIDIA Nemotron 3 Nano 4B오픈NVIDIA	8.6	—	—	—
30	▼ 2	Llama 3.1 Nemotron Nano 4B v1.1 (Reasoning)오픈NVIDIA	8.4	—	—	—
31	▼ 2	Llama 3.1 Instruct 405B오픈Meta	8.3	—	—	—
32	▼ 2	LFM2.5-8B-A1B오픈Liquid AI	8.1	$0.000	336	2.0초
33	▼ 1	Llama 3.1 Nemotron Instruct 70B오픈NVIDIA	7.4	$1.2	126	3.5초
34	▼ 3	Llama 3.1 Instruct 8B오픈Meta	7.4	$0.028	136	0.9초
35	▼ 2	Gemma 3 27B Instruct오픈Google	7.4	$0.000	—	—
36	▼ 2	Llama 3.1 Instruct 70B오픈Meta	6.5	$0.56	56	1.4초
37	▼ 2	Granite 4.1 8B오픈IBM	6.4	$0.063	112	0.8초
38	▼ 1	R1 1776오픈Perplexity	6.0	—	—	—
39	▼ 1	Llama 3.2 Instruct 90B (Vision)오픈Meta	6.0	—	—	—
40	▼ 4	Phi-4 Mini Instruct오픈Microsoft	5.7	$0.000	44	0.9초
41	▼ 2	Gemma 3 12B Instruct오픈Google	5.5	$0.000	—	—
42	▼ 2	Granite 4.0 H Small오픈IBM	4.9	$0.11	22	24초
43	▼ 1	LFM2 24B A2B오픈Liquid AI	4.6	—	—	—
44	▼ 1	Phi-4오픈Microsoft	4.6	$0.22	25	3.1초
45	▼ 4	Granite 4.1 3B오픈IBM	4.4	—	—	—
46	▼ 2	Phi-3 Mini Instruct 3.8B오픈Microsoft	4.3	—	—	—
47	▼ 2	Gemma 3n E4B Instruct Preview (May '25)오픈Google	4.2	—	—	—
48	▼ 2	Phi-4 Multimodal Instruct오픈Microsoft	4.2	$0.000	18	0.8초
49	▼ 2	Llama 2 Chat 7B오픈Meta	3.9	$0.10	—	—
50	▼ 2	Llama 3.2 Instruct 3B오픈Meta	3.9	—	—	—
51	▼ 2	Llama 3 Instruct 70B오픈Meta	3.1	$1.2	—	—
52	▼ 2	Arctic Instruct오픈Snowflake	3.0	—	—	—
53	▼ 2	Llama 3.2 Instruct 11B (Vision)오픈Meta	3.0	$0.35	48	0.9초
54	▼ 2	Llama 2 Chat 70B오픈Meta	2.6	—	—	—
55	▼ 2	Llama 2 Chat 13B오픈Meta	2.6	—	—	—
56	▼ 2	DBRX Instruct오픈Databricks	2.6	—	—	—
57	▼ 2	LFM2.5-1.2B-Thinking오픈Liquid AI	2.3	—	—	—
58	▼ 2	LFM2 2.6B오픈Liquid AI	2.3	—	—	—
59	▼ 2	LFM2.5-1.2B-Instruct오픈Liquid AI	2.3	—	—	—
60	▼ 2	Granite 4.0 H 1B오픈IBM	2.2	—	—	—
61	▼ 2	Gemma 3 270M오픈Google	2.0	—	—	—
62	▼ 2	Granite 4.0 Micro오픈IBM	2.0	—	—	—
63	▼ 2	Llama 65B오픈Meta	1.7	—	—	—
64	▼ 2	Granite 4.0 1B오픈IBM	1.6	—	—	—
65	▼ 2	LFM2 8B A1B오픈Liquid AI	1.3	—	—	—
66	▼ 2	Granite 3.3 8B (Non-reasoning)오픈IBM	1.3	$0.085	15	27초
67	▼ 2	Gemma 3 1B Instruct오픈Google	1.0	$0.000	—	—
67	▼ 2	Gemma 3 4B Instruct오픈Google	1.0	$0.000	—	—
67	▼ 2	Gemma 3n E2B Instruct오픈Google	1.0	$0.000	—	—
67	▼ 2	Gemma 3n E4B Instruct오픈Google	1.0	$0.075	57	1.2초
67	▼ 2	Granite 4.0 350M오픈IBM	1.0	—	—	—
67	▼ 2	Granite 4.0 H 350M오픈IBM	1.0	—	—	—
67	▼ 2	LFM2 1.2B오픈Liquid AI	1.0	—	—	—
67	▼ 2	LFM2.5-VL-1.6B오픈Liquid AI	1.0	$0.000	334	7.4초
67	▼ 2	Llama 3 Instruct 8B오픈Meta	1.0	$0.070	—	—
67	▼ 2	Llama 3.2 Instruct 1B오픈Meta	1.0	—	—	—

모델 26개 더 보기

지능의 발전

각 개발사가 그 시점에 가진 최고 모델의 Artificial Analysis Intelligence Index를 모델 공개 월 기준으로 보여 줍니다.

가장 많이 오름 · Anthropic +57.2현재 최고 · Anthropic 64.6최고와 최저의 차이 18.1점기간 2025년 11월 – 2026년 11월

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