DW AI LAB

공식 자료 기반의 AI 분석

분석 방법

벤치마크 상세

LiveBench

최신 종합 평가: 최신 모델의 수학, 코딩, 추론, 언어 이해 흐름을 빠르게 볼 때

공식 리더보드 보기 공식 LiveBench 방식에 맞춰 분야별 평균을 먼저 계산하고, 공개된 분야 평균을 다시 동일 가중치로 평균했습니다. 해당 릴리스에 공개된 모델 전체를 담습니다.

모든 공개 결과

모델 검색 · 점수순
확인 2026. 10. 13.2026-06-25 releasemean of 7 category averages
70개 결과원문 ↗
순서모델 / 실행 구성점수평가 조건·근거
1claude-fable-5-1-max-effort83.412026-06-25 release
2claude-opus-5-5-max-effort83.222026-06-25 release
3claude-fable-5-max-effort82.972026-06-25 release
4gpt-6-astra-max82.162026-06-25 release
5claude-opus-5-5-xhigh-effort82.062026-06-25 release
6gpt-6.1-sol-max81.622026-06-25 release
7muse-spark-1.3-xhigh81.592026-06-25 release
8muse-wonder-1.0-max81.162026-06-25 release
9deepseek-v4.1-flash-max81.112026-06-25 release
10gpt-6.1-sol-xhigh81.112026-06-25 release
11gpt-5.6-sol-max81.052026-06-25 release
12gpt-5.5-xhigh80.192026-06-25 release
13claude-opus-5-max-effort80.082026-06-25 release
14smaug-agentic79.512026-06-25 release
15gpt-6-sol-max79.252026-06-25 release
16kimi-k379.192026-06-25 release
17gemini-3.7-flash-high78.832026-06-25 release
18qwen3.8-max78.462026-06-25 release
19grok-4.678.042026-06-25 release
20gpt-5.4-xhigh77.972026-06-25 release
21muse-spark-1.2-xhigh77.952026-06-25 release
22gpt-5.6-terra-max77.942026-06-25 release
23claude-sonnet-5-5-xhigh-effort77.752026-06-25 release
24deepseek-v4-pro-081377.442026-06-25 release
25smaug-flash77.442026-06-25 release
26grok-4.7-xhigh77.42026-06-25 release
27gemini-3.1-pro-preview-high76.952026-06-25 release
28smaug-mini76.932026-06-25 release
29deepseek-v4-flash-vision-exp76.762026-06-25 release
30claude-opus-4-7-xhigh-effort76.532026-06-25 release
31claude-opus-4-8-max-effort76.222026-06-25 release
32qwen3.8-flash-next76.192026-06-25 release
33glm-5.376.142026-06-25 release
34claude-sonnet-5-xhigh-effort76.042026-06-25 release
35gemini-3.8-flash-high75.832026-06-25 release
36grok-4.575.772026-06-25 release
37claude-sonnet-5-5-max-effort75.672026-06-25 release
38muse-spark-1.1-xhigh75.32026-06-25 release
39qwen3.8-27b75.272026-06-25 release
40gemini-3.5-flash-high74.642026-06-25 release
41gpt-5.2-2025-12-11-high74.632026-06-25 release
42claude-opus-4-6-thinking-auto-high-effort74.522026-06-25 release
43deepseek-v4-flash-073174.172026-06-25 release
44gpt-5.2-codex73.982026-06-25 release
45gemini-3.6-flash-high73.592026-06-25 release
46gpt-5.6-luna-max73.562026-06-25 release
47glm-5.273.162026-06-25 release
48qwen3.7-max73.142026-06-25 release
49claude-sonnet-4-6-thinking-auto-medium-effort72.992026-06-25 release
50claude-opus-4-5-20251101-thinking-64k-high-effort72.582026-06-25 release
51claude-haiku-5-5-xhigh-effort72.072026-06-25 release
52gpt-6-luna-max72.032026-06-25 release
53inkling-xhigh71.922026-06-25 release
54mistral-large-4-high71.842026-06-25 release
55glm-5.3-flash71.592026-06-25 release
56deepseek-v4-pro71.572026-06-25 release
57kimi-k2.6-thinking70.542026-06-25 release
58gpt-5.4-nano-xhigh69.582026-06-25 release
59ox-alpha-max69.242026-06-25 release
60qwen3.6-plus68.912026-06-25 release
61kimi-k2.7-code68.412026-06-25 release
62claude-haiku-5-5-max-effort67.882026-06-25 release
63grok-build-0.167.782026-06-25 release
64nemotron-3-ultra-550b-a55b67.362026-06-25 release
65minimax-m367.262026-06-25 release
66gpt-5.4-mini-xhigh66.372026-06-25 release
67deepseek-v4-flash65.482026-06-25 release
68qwen3.6-27b64.032026-06-25 release
69gemini-3.5-flash-lite-high63.942026-06-25 release
70grok-4.362.252026-06-25 release

공개된 숫자만 높은 순으로 표시합니다. 순서가 신뢰구간을 고려한 확정 순위는 아닙니다. 공식 LiveBench 방식에 맞춰 분야별 평균을 먼저 계산하고, 공개된 분야 평균을 다시 동일 가중치로 평균했습니다. 해당 릴리스에 공개된 모델 전체를 담습니다.

평가 방법 확인 ↗

읽는 법

전체 벤치마크

오염을 줄이기 위해 새 문제를 계속 추가하는 최신 LLM 종합 리더보드입니다.

고정된 오래된 시험보다 최신 모델 간 상대적 흐름을 보기에 좋습니다. 세부 task별 점수를 함께 봐야 합니다.