AI 모델 평가

AI 모델 전체 비교표

← AI 모델 평가로 돌아가기

2026년 7월 기준 ai.caston.blog에 등록된 50개 AI 모델을 종합점수 순으로 정리한 비교표입니다. 각 모델명을 클릭하면 상세 리뷰로 이동합니다.

순위 모델 제공사 종합 코딩 라이선스 로컬실행 가격 (1M 토큰) 컨텍스트
1 GPT-5.5 OpenAI 94 91 Closed ❌ $5.00 / $30.00 922K
2 o4 OpenAI 93 90 Closed ❌ $8.00 / $32.00 200K
3 Claude Opus 4.8 Anthropic 92 93 Closed ❌ $5 / $25 1M
4 Gemini 3 Pro Google 91 88 Closed ❌ $2.50 / $12.00 2M
5 GLM-5.2 Z AI 90 89 MIT (Open) ✅ $0.90 (blended) 1M
6 Llama 4 Behemoth Meta 90 85 Llama 4 Community ❌ 오픈웨이트(API 전용) 1M
7 Grok 4.20 xAI 89 85 Closed ❌ $6.00 / $24.00 512K
8 DeepSeek R2 DeepSeek 89 90 MIT (Open) ✅ $0.55 (blended) 128K
9 Claude Sonnet 4.6 Anthropic 88 87 Closed ❌ $3.00 / $15.00 1M
10 DeepSeek V4 Pro DeepSeek 87 86 MIT (Open) ✅ $0.18 (blended) 1M
11 DeepSeek V3 DeepSeek 87 91 MIT (Open) ✅ $0.24 / $0.90 164K
12 Grok 4.3 xAI 86 82 Closed ❌ $4.00 / $16.00 256K
13 Qwen3.7 Max Alibaba 86 83 Closed (API) ❌ $1.43 (blended) 1M
14 Claude Sonnet 4.5 Anthropic 86 85 Closed ❌ $3.00 / $15.00 500K
15 Kimi K2.6 Moonshot AI 85 88 Modified MIT (Open) ✅ $0.70 (blended) 256K
16 Gemini 2.5 Pro Google 85 86 Closed ❌ $1.25 / $10.00 1M
17 DeepSeek Coder V3 DeepSeek 84 92 MIT (Open) ✅ $0.30 / $1.20 128K
18 Mistral Large 3 2512 Mistral 84 79 Apache 2.0 (Open) ✅ $2.00 / $6.00 256K
19 Llama 4 Maverick Meta 83 80 Llama 4 Community ✅ 오픈웨이트 1M
20 MiniMax-M3 MiniMax 83 80 Apache 2.0 (Open) ✅ $0.22 (blended) 1M
21 GLM-4.6 Z AI 83 80 MIT (Open) ✅ $0.60 (blended) 200K
22 Gemini 3.5 Flash Google 82 79 Closed ❌ $1.50 / $9.00 1M
23 Qwen3 Coder Alibaba 82 90 Apache 2.0 (Open) ✅ $0.40 (blended) 256K
24 Qwen2.5 Max Alibaba 81 78 Closed (API) ❌ $1.60 (blended) 128K
25 Nemotron 4 340B Nvidia 81 76 NVIDIA Open License ✅ 오픈웨이트 128K
26 Kimi K2 0711 Moonshot AI 80 83 Modified MIT (Open) ✅ $0.60 / $2.50 128K
27 DeepSeek V4 Flash DeepSeek 80 78 MIT (Open) ✅ $0.14 / $0.28 1M
28 Qwen3 235B A22B Alibaba 80 82 Apache 2.0 (Open) ✅ 오픈웨이트 262K
29 Grok 3 xAI 80 77 Closed ❌ $3.00 / $15.00 128K
30 Nova Pro Amazon 80 75 Closed (API) ❌ $0.80 / $3.20 300K
31 Sonar Large Perplexity 79 70 Closed (API) ❌ $1.00 / $1.00 128K
32 Llama 4 Scout Meta 79 74 Llama 4 Community ✅ 오픈웨이트 10M
33 MiniMax M1 MiniMax 78 75 Apache 2.0 (Open) ✅ $0.55 / $2.20 1M
34 Gemini 2.5 Flash Google 78 76 Closed ❌ $0.30 / $2.50 1M
35 GPT-4o OpenAI 78 75 Closed ❌ $2.50 / $10.00 128K
36 Codestral 3 Mistral 78 85 Mistral Non-Prod (제한적 오픈) ✅ $0.30 / $0.90 256K
37 Command A Cohere 77 72 Closed (API) ❌ $2.50 / $10.00 256K
38 GPT-5.5 Mini OpenAI 76 73 Closed ❌ $0.30 / $1.20 400K
39 Kimi K1.5 Moonshot AI 76 74 Modified MIT (Open) ✅ $0.50 / $2.00 200K
40 Jamba 2 Large AI21 75 70 Jamba Open License ✅ $0.50 / $1.50 256K
41 Claude Haiku 4.5 Anthropic 74 70 Closed ❌ $0.80 / $4.00 200K
42 Gemma 3 27B Google 74 70 Gemma License (Open) ✅ 오픈웨이트 128K
43 MiniMax Text-01 MiniMax 74 70 Apache 2.0 (Open) ✅ $0.20 / $1.10 4M
44 Llama 3.3 70B Meta 72 68 Llama 3.3 Community ✅ 오픈웨이트 128K
45 Mistral Small 3 Mistral 71 68 Apache 2.0 (Open) ✅ $0.10 / $0.30 128K
46 Granite 4 IBM 70 68 Apache 2.0 (Open) ✅ 오픈웨이트 128K
47 Orca 3 Microsoft 69 64 MIT (Open) ✅ 오픈웨이트 32K
48 Phi-4 Microsoft 68 65 MIT (Open) ✅ 오픈웨이트 16K
49 GPT-4o-mini OpenAI 65 62 Closed ❌ $0.15 / $0.60 128K
50 Phi-4-mini Microsoft 60 56 MIT (Open) ✅ 오픈웨이트 8K

※ 종합·코딩 점수는 자체 평가 기준(0~100)이며, 가격은 1M 토큰당 입력/출력 기준입니다. 데이터 기준일: 2026년 7월 4일.