AI 벤치마크 비교
종합지능 투표 기준 실시간 성능 랭킹 텍스트 / 코딩 / 비전 / 이미지 / 영상
종합지능 투표 기준 실제 사용자 배틀 기반 ELO 점수 체크박스로 모델 선택 후 비교 가능
모델 2-4개 선택 비교하기
| 순위 | 모델명 (정확한 버전) | 제공사 | Text ELO | Code ELO | Vision ELO | 투표수 | 컨텍스트 | 출시일 | |
|---|---|---|---|---|---|---|---|---|---|
| 1 | Claude Opus 4.6 Thinking | Anthropic | 1502 | 1546 | 64,805 | 200K | 2026-03 | ||
| 2 | Claude Opus 4.6 | Anthropic | 1496 | 1548 | 68,516 | 200K | 2026-03 | ||
| 3 | Gemini 3.1 Pro | 1493 | 1454 | 1276 | 14,677 | 2,000K | 2026-03 | ||
| 4 | Grok 4.20 Beta | xAI | 1492 | 7,396 | 131K | 2026-03 | |||
| 5 | claude-opus-4-7-thinking | Anthropic | 1489 | 52,303 | 2026-07 | ||||
| 6 | GPT-5.2 | OpenAI | 1482 | 1275 | 10,140 | 128K | 2026-02 | ||
| 7 | gemini-3.5-flash | 1482 | 15,261 | 1,048K | 2026-07 | ||||
| 8 | claude-opus-4-7 | Anthropic | 1481 | 53,428 | 2026-07 | ||||
| 9 | Grok 4.20 Reasoning | xAI | 1481 | 4,504 | 131K | 2026-03 | |||
| 10 | gemini-3.5-flash-high | 1480 | 10,011 | 2026-08 | |||||
| 11 | Gemini 3 Pro | 1479 | 1290 | 41,242 | 2,000K | 2026-01 | |||
| 12 | gemini-3.1-pro-preview | 1479 | 86,683 | 1,048K | 2026-07 | ||||
| 13 | gemini-3.6-flash | 1477 | 6,065 | 1,048K | 2026-08 | ||||
| 14 | Claude Opus 4.5 Thinking | Anthropic | 1474 | 1489 | 37,036 | 200K | 2025-11 | ||
| 15 | gemini-3.5-flash-medium | 1472 | 16,248 | 2026-08 | |||||
| 16 | Grok 4.1 Thinking | xAI | 1470 | 5,200 | 131K | 2026-02 | |||
| 17 | qwen3.5-max-preview | Alibaba | 1470 | 21,306 | 2026-07 | ||||
| 18 | gpt-5.4-high | OpenAI | 1469 | 60,292 | 2026-07 | ||||
| 19 | gpt-5.5-high | OpenAI | 1469 | 47,572 | 2026-07 | ||||
| 20 | Gemini 3 Flash | 1466 | 1274 | 30,643 | 1,000K | 2026-01 | |||
| 21 | Claude Opus 4.5 | Anthropic | 1465 | 1465 | 13,313 | 200K | 2025-11 | ||
| 22 | GPT-5.5 | OpenAI | 1464 | 48,792 | 1,050K | 2026-04 | |||
| 23 | claude-opus-4-8-thinking | Anthropic | 1462 | 32,817 | 2026-07 | ||||
| 24 | Qwen 3.5 Max | Alibaba | 1460 | 7,800 | 131K | 2026-02 | |||
| 25 | GPT-5.3 | OpenAI | 1458 | 8,900 | 128K | 2026-02 | |||
| 26 | gemini-2.5-pro | 1457 | 124,076 | 1,048K | 2026-07 | ||||
| 27 | Claude Sonnet 4.6 | Anthropic | 1457 | 1521 | 58,826 | 200K | 2026-03 | ||
| 28 | Gemini 3 Flash Thinking | 1455 | 1258 | 11,942 | 1,000K | 2026-01 | |||
| 29 | gpt-5.6-sol-xhigh | OpenAI | 1455 | 8,359 | 2026-08 | ||||
| 30 | claude-opus-4-8 | Anthropic | 1453 | 33,401 | 2026-07 | ||||
| 31 | GPT-5.4 | OpenAI | 1451 | 1457 | 63,194 | 128K | 2026-03 | ||
| 32 | grok-4.20-beta-0309-reasoning | xAI | 1451 | 61,762 | 2026-07 | ||||
| 33 | grok-4.5 | xAI | 1451 | 9,998 | 500K | 2026-08 | |||
| 34 | GPT-5.4 Standard | OpenAI | 1450 | 6,000 | 128K | 2026-03 | |||
| 35 | grok-4.20-multi-agent-beta-0309 | xAI | 1450 | 60,452 | 2026-07 | ||||
| 36 | claude-opus-4-5-20251101 | Anthropic | 1449 | 70,573 | 2026-07 | ||||
| 37 | gpt-5.6-terra-xhigh | OpenAI | 1448 | 8,621 | 2026-08 | ||||
| 38 | claude-opus-4-5-20251101-thinking-32k | Anthropic | 1446 | 36,980 | 2026-07 | ||||
| 39 | Grok 4.1 | xAI | 1445 | 7,200 | 131K | 2026-02 | |||
| 40 | GLM-5 | Zhipu AI | 1445 | 1445 | 27,644 | 128K | 2026-01 | ||
| 41 | grok-4.20-beta1 | xAI | 1444 | 26,598 | 2026-07 | ||||
| 42 | gemini-3-flash (thinking-minimal) | 1442 | 85,621 | 2026-07 | |||||
| 43 | gpt-5.1-high | OpenAI | 1441 | 40,700 | 2026-07 | ||||
| 44 | gpt-5.2-chat-latest-20260210 | OpenAI | 1438 | 34,060 | 2026-07 | ||||
| 45 | gemini-3.5-flash-lite | 1438 | 6,015 | 1,048K | 2026-08 | ||||
| 46 | claude-sonnet-4-5-20250929 | Anthropic | 1437 | 80,347 | 2026-07 | ||||
| 47 | grok-4.1-thinking | xAI | 1437 | 65,116 | 2026-07 | ||||
| 48 | grok-4.1 | xAI | 1436 | 67,322 | 2026-07 | ||||
| 49 | claude-sonnet-4-5-20250929-thinking-32k | Anthropic | 1432 | 81,949 | 2026-07 | ||||
| 50 | Kimi K2.5 Thinking | Moonshot AI | 1430 | 1246 | 7,605 | 200K | 2026-02 | ||
| 51 | gpt-5.6-luna-xhigh | OpenAI | 1429 | 8,890 | 2026-08 | ||||
| 52 | Gemini 2.5 Pro | 1425 | 1247 | 83,351 | 2,000K | 2025-03 | |||
| 53 | grok-3-preview-02-24 | xAI | 1425 | 32,863 | 2026-07 | ||||
| 54 | GPT-5.1 | OpenAI | 1422 | 1248 | 43,332 | 128K | 2026-01 | ||
| 55 | Claude Sonnet 4.5 | Anthropic | 1420 | 6,375 | 200K | 2025-09 | |||
| 56 | gpt-5.5-instant | OpenAI | 1419 | 25,708 | 2026-07 | ||||
| 57 | claude-opus-4-1-20250805-thinking-16k | Anthropic | 1418 | 49,741 | 2026-07 | ||||
| 58 | claude-opus-4-1-20250805 | Anthropic | 1417 | 77,292 | 2026-07 | ||||
| 59 | gpt-4.5-preview-2025-02-27 | OpenAI | 1417 | 14,547 | 2026-07 | ||||
| 60 | gemini-2.5-flash | 1417 | 124,477 | 1,048K | 2026-07 | ||||
| 61 | gpt-5.2-high | OpenAI | 1416 | 48,052 | 2026-07 | ||||
| 62 | gemini-3.1-flash-lite-preview | 1415 | 57,809 | 1,048K | 2026-07 | ||||
| 63 | gpt-5.4-mini-high | OpenAI | 1413 | 48,963 | 2026-07 | ||||
| 64 | ERNIE 5.0 | Baidu | 1410 | 4,200 | 128K | 2026-01 | |||
| 65 | GPT-5.4 Nano | OpenAI | 400K | 2026-03 | |||||
| 66 | GPT-5.4 Mini | OpenAI | 400K | 2026-03 | |||||
| 67 | GPT-5.4 Pro | OpenAI | 1,050K | 2026-03 | |||||
| 68 | GPT-5.3 Chat | OpenAI | 128K | 2026-03 | |||||
| 69 | GPT-5.3-Codex | OpenAI | 400K | 2026-03 | |||||
| 70 | GPT-5.2-Codex | OpenAI | 400K | 2026-03 | |||||
| 71 | GPT-5.2 Chat | OpenAI | 128K | 2026-03 | |||||
| 72 | GPT-5.2 Pro | OpenAI | 400K | 2026-03 | |||||
| 73 | GPT-5.1-Codex-Max | OpenAI | 400K | 2026-03 | |||||
| 74 | GPT-5.1 Chat | OpenAI | 128K | 2026-03 | |||||
| 75 | GPT-5.1-Codex | OpenAI | 400K | 2026-03 | |||||
| 76 | GPT-5.1-Codex-Mini | OpenAI | 400K | 2026-03 | |||||
| 77 | GPT-5 Image Mini | OpenAI | 400K | 2026-03 | |||||
| 78 | GPT-5 Image | OpenAI | 400K | 2026-03 | |||||
| 79 | o3 Deep Research | OpenAI | 200K | 2026-03 | |||||
| 80 | o4 Mini Deep Research | OpenAI | 200K | 2026-03 | |||||
| 81 | GPT-5 Pro | OpenAI | 400K | 2026-03 | |||||
| 82 | GPT-5 Codex | OpenAI | 400K | 2026-03 | |||||
| 83 | GPT-5 Chat | OpenAI | 128K | 2026-03 | |||||
| 84 | GPT-5 | OpenAI | 400K | 2026-03 | |||||
| 85 | GPT-5 Mini | OpenAI | 400K | 2026-03 | |||||
| 86 | GPT-5 Nano | OpenAI | 400K | 2026-03 | |||||
| 87 | o3 Pro | OpenAI | 200K | 2026-03 | |||||
| 88 | o4 Mini High | OpenAI | 200K | 2026-03 | |||||
| 89 | o3 | OpenAI | 200K | 2026-03 | |||||
| 90 | o4 Mini | OpenAI | 200K | 2026-03 | |||||
| 91 | o3 Mini High | OpenAI | 200K | 2026-03 | |||||
| 92 | o3 Mini | OpenAI | 200K | 2026-03 | |||||
| 93 | GPT-5.4 Image 2 | OpenAI | 272K | 2026-04 | |||||
| 94 | GPT-5.5 Pro | OpenAI | 1,050K | 2026-04 | |||||
| 95 | GPT-5.6 Luna Pro | OpenAI | 1,050K | 2026-08 | |||||
| 96 | GPT-5.6 Luna | OpenAI | 1,050K | 2026-08 | |||||
| 97 | GPT-5.6 Terra Pro | OpenAI | 1,050K | 2026-08 | |||||
| 98 | GPT-5.6 Terra | OpenAI | 1,050K | 2026-08 | |||||
| 99 | GPT-5.6 Sol Pro | OpenAI | 1,050K | 2026-08 | |||||
| 100 | GPT-5.6 Sol | OpenAI | 1,050K | 2026-08 |
오픈소스 LLM 랭킹
(HuggingFace Open LLM Leaderboard 별도 분류)
| 순위 | 모델명 | 제공사 | Arena ELO | MMLU | 컨텍스트 | 출시일 | 바로가기 |
|---|---|---|---|---|---|---|---|
| 1 | mistral-large-3 | Mistral AI | 1429 | 2026-07 | HF | ||
| 2 | deepseek-v3.2 | DeepSeek | 1424 | 131K | 2026-07 | HF | |
| 3 | deepseek-v3.2-exp-thinking | DeepSeek | 1424 | 2026-07 | HF | ||
| 4 | deepseek-v3.2-exp | DeepSeek | 1422 | 163K | 2026-07 | HF | |
| 5 | deepseek-v3.1-terminus-thinking | DeepSeek | 1419 | 2026-07 | HF | ||
| 6 | deepseek-v3.2-thinking | DeepSeek | 1419 | 2026-07 | HF | ||
| 7 | deepseek-v3.1 | DeepSeek | 1419 | 163K | 2026-07 | HF | |
| 8 | deepseek-v3.1-terminus | DeepSeek | 1417 | 163K | 2026-07 | HF | |
| 9 | deepseek-v3.1-thinking | DeepSeek | 1417 | 2026-07 | HF | ||
| 10 | Qwen 3.5 397B | Alibaba | 1405 | 32K | 2026-02 | HF | |
| 11 | DeepSeek R2 | DeepSeek | 1390 | 88.0 | 128K | 2025-12 | HF |
| 12 | Qwen 3 235B-A22B | Alibaba | 1380 | 87.8 | 128K | 2025-04 | HF |
| 13 | DeepSeek V3 | DeepSeek | 1370 | 88.5 | 128K | 2024-12 | HF |
| 14 | Llama 4 Maverick | Meta | 1350 | 87.5 | 1,000K | 2025-04 | HF |
| 15 | Llama 4 Scout | Meta | 1320 | 84.0 | 10,000K | 2025-04 | HF |
| 16 | Mistral Large 3 | Mistral AI | 1310 | 84.0 | 128K | 2025-07 | HF |
코딩 전용 종합지능 투표 Code Arena 기준
| 순위 | 모델명 | 제공사 | Code ELO | Text ELO | 출시일 |
|---|---|---|---|---|---|
| 1 | Claude Opus 4.6 | Anthropic | 1548 | 1496 | 2026-03 |
| 2 | Claude Opus 4.6 Thinking | Anthropic | 1546 | 1502 | 2026-03 |
| 3 | Claude Sonnet 4.6 | Anthropic | 1521 | 1457 | 2026-03 |
| 4 | Claude Opus 4.5 Thinking | Anthropic | 1489 | 1474 | 2025-11 |
| 5 | Claude Opus 4.5 | Anthropic | 1465 | 1465 | 2025-11 |
| 6 | GPT-5.4 | OpenAI | 1457 | 1451 | 2026-03 |
| 7 | Gemini 3.1 Pro | 1454 | 1493 | 2026-03 | |
| 8 | GLM-5 | Zhipu AI | 1445 | 1445 | 2026-01 |
이미지 이해 / 멀티모달 종합지능 투표 Vision Arena 기준
| 순위 | 모델명 | 제공사 | Vision ELO | Text ELO | 출시일 |
|---|---|---|---|---|---|
| 1 | Gemini 3 Pro | 1290 | 1479 | 2026-01 | |
| 2 | Gemini 3.1 Pro | 1276 | 1493 | 2026-03 | |
| 3 | GPT-5.2 | OpenAI | 1275 | 1482 | 2026-02 |
| 4 | Gemini 3 Flash | 1274 | 1466 | 2026-01 | |
| 5 | Gemini 3 Flash Thinking | 1258 | 1455 | 2026-01 | |
| 6 | GPT-5.1 | OpenAI | 1248 | 1422 | 2026-01 |
| 7 | Gemini 2.5 Pro | 1247 | 1425 | 2025-03 | |
| 8 | Kimi K2.5 Thinking | Moonshot AI | 1246 | 1430 | 2026-02 |
이미지 생성 종합지능 투표 Text-to-Image Arena 기준
| 순위 | 모델명 | 제공사 | Image Gen ELO | 투표수 | 출시일 |
|---|---|---|---|---|---|
| 1 | Gemini 3.1 Flash Image | 1266 | 15,317 | 2026-03 | |
| 2 | GPT Image 1.5 | OpenAI | 1244 | 62,556 | 2026-02 |
| 3 | Gemini 3 Pro Image | 1235 | 58,168 | 2026-01 | |
| 4 | Grok Imagine | xAI | 1173 | 48,767 | 2025-12 |
| 5 | FLUX 2 Max | Black Forest Labs | 1167 | 66,012 | 2026-01 |
영상 생성 종합지능 투표 Text-to-Video Arena 기준
| 순위 | 모델명 | 제공사 | Video ELO | 투표수 | 출시일 |
|---|---|---|---|---|---|
| 1 | Veo 3.1 Audio (1080p) | 1381 | 5,537 | 2026-03 | |
| 2 | Sora 2 Pro | OpenAI | 1367 | 18,963 | 2026-01 |
| 3 | Grok Imagine Video | xAI | 1358 | 33,739 | 2025-12 |
평가 방식
- lmarena.ai Chatbot Arena Leaderboard (https://lmarena.ai/leaderboard)
- HuggingFace Open LLM Leaderboard v2 (open-source 모델 별도 분류)
- OpenRouter API (신규 모델 자동 감지)
새 모델 출시 시 자동으로 추가됩니다. 실제 성능은 사용 환경에 따라 다를 수 있습니다.