본문으로 건너뛰기
AXyNowAX IS NOW
사업 운영EP11

📜법무

Legal (contracts·family)

외피 — 산업 도메인
법무 (계약·가족·상속)
내용 — 측정하는 AI 능력
  • · 한국 법령 인용 정확도
  • · 사실관계 추론·위험 식별
  • · 계약 조항·민법 구조화

모델별 종합 점수

✓ 챗봇 1턴

측정일 2026-06-05T02:40:47+00:00 · 5개 항목 × 100점 기준

채점자 editor · max_tokens 32768 · temp 0.7 · attempts 3 · reasoning_effort medium

모델
1ClaudeClaude Sonnet 4.6
5/5918586919189.4
2ClaudeClaude Opus 4.8
5/5888787928989.2
3OpenAIGPT-5.5
5/5888486889087.8
4MiniMaxMiniMax M3
5/5849187829486.0
5DeepSeekDeepSeek V4 Pro
5/5878383878585.6
6Google GeminiGemini 3.1 Pro
5/5868383878485.4
7Google GeminiGemini 3.5 Flash
5/5868282868585.0
8XiaomiMimo V2.5 Pro
5/5858282858785.0
9QWenQwen 3.7 Max
5/5858383868585.0
10Moonshot AIKimi K2.6
5/5838382838583.4
11OpenAIGPT-5.4 Mini
5/5848182848283.0
12QWenQwen 3.7 Plus
5/5808080808881.2
13Google GeminiGemma 4 12B
5/5788582788580.4
14DeepSeekDeepSeek V4 Flash
5/5817977818080.2
15xAIGrok 4.3
5/5817879818080.0
16GLM 5.1
5/5788080798279.6
17StepFunStep 3.7 Flash
5/5778882748879.6
18Google GeminiGemini 3.1 Flash Lite
5/5807878807679.0
19Google GeminiGemma 4 26B A4B
5/5807878807979.0
20Google GeminiGemma 4 31B
5/5807777807978.8
21QWenQwen 3.6 35B A3B
5/5717774728073.4
22QWenQwen 3.6 27B
5/5717674717773.0
23LG AIEXAONE 4.5 33B
5/56080607410073.0
24NVIDIANemotron 3 Ultra 550B
5/5588365669169.6
25NaverHyperCLOVAX SEED Think 32B
5/5608060648066.4
26Solar Pro 3
5/5467460609262.8
27QWenQwen 3.5 9B
5/5528264567961.8
28Mistral AIMistral Small 4
5/5536457526456.0
29Google GeminiGemma 4 E2B
5/5455342415445.2
30KakaoKanana 2 30B-A3B Thinking
5/5325050386042.6
31NaverHyperCLOVAX SEED 1.5B
5/5344633304335.2
32Liquid AILFM2.5 8B-A1B
5/5294527254531.2

문항별 점수

5 문항

각 문항당 모델 세부 점수. 응답 원문·근거는 문항 카드 우측 링크.

법무 · 문항 1공급계약서 — 위험 조항 식별 + 수정 제안공개

공급계약서 — 위험 조항 식별 + 수정 제안

본문·raw·근거 →
모델
정확성의도 파악신중함한국 맥락짜임새avg
Claude Sonnet 4.6Anthropic
958085959592
Claude Opus 4.8Anthropic
908585959090
GPT-5.5OpenAI
908285909289
MiniMax M3Minimax
909288909591
DeepSeek V4 ProDeepSeek
888080888586
Gemini 3.1 ProGoogle
888082888285
Gemini 3.5 FlashGoogle
887880888585
Mimo V2.5 ProXiaomi
857880858584
Qwen 3.7 MaxAlibaba
858080868284
Kimi K2.6Moonshot
868082868585
GPT-5.4 MiniOpenAI
857880858283
Qwen 3.7 PlusAlibaba
8080808010083
Gemma 4 12BGoogle
808482808682
DeepSeek V4 FlashDeepSeek
857575858082
Grok 4.3xAI
827578827880
GLM 5.1Z.ai
807880828281
Step 3.7 FlashStepFun
859085829085
Gemini 3.1 Flash LiteGoogle
827575827579
Gemma 4 26B A4BGoogle
787272787576
Gemma 4 31BGoogle
787272767575
Qwen 3.6 35B A3BAlibaba
827880828281
Qwen 3.6 27BAlibaba
727575747874
EXAONE 4.5 33BLG AI
6080608010075
Nemotron 3 Ultra 550BNVIDIA
608270809576
HyperCLOVAX SEED Think 32BNaver
608060608065
Solar Pro 3Upstage
407060608059
Qwen 3.5 9BAlibaba
558270587063
Mistral Small 4Mistral
707072706870
Gemma 4 E2BGoogle
465343415646
Kanana 2 30B-A3B ThinkingKakao
406050406046
HyperCLOVAX SEED 1.5BNaver
324432274133
LFM2.5 8B-A1BLiquid AI
384936365240
법무 · 문항 2이혼 — 재산분할 청구 + 양육비 산정비공개

이혼·재산분할

본문·raw·근거 →
모델
정확성의도 파악신중함한국 맥락짜임새avg
Claude Sonnet 4.6Anthropic
908282909088
Claude Opus 4.8Anthropic
958888959092
GPT-5.5OpenAI
888282889087
MiniMax M3Minimax
729085709278
DeepSeek V4 ProDeepSeek
858080858283
Gemini 3.1 ProGoogle
908282908287
Gemini 3.5 FlashGoogle
908282908587
Mimo V2.5 ProXiaomi
908583909088
Qwen 3.7 MaxAlibaba
838080848282
Kimi K2.6Moonshot
908585928889
GPT-5.4 MiniOpenAI
858080858283
Qwen 3.7 PlusAlibaba
808080808080
Gemma 4 12BGoogle
768480768479
DeepSeek V4 FlashDeepSeek
827878828081
Grok 4.3xAI
767575767876
GLM 5.1Z.ai
808080808280
Step 3.7 FlashStepFun
788882758880
Gemini 3.1 Flash LiteGoogle
787575787577
Gemma 4 26B A4BGoogle
827878827880
Gemma 4 31BGoogle
827878827880
Qwen 3.6 35B A3BAlibaba
828080848282
Qwen 3.6 27BAlibaba
667470667469
EXAONE 4.5 33BLG AI
6080608010075
Nemotron 3 Ultra 550BNVIDIA
608470729073
HyperCLOVAX SEED Think 32BNaver
608060608065
Solar Pro 3Upstage
5080606010066
Qwen 3.5 9BAlibaba
628572708071
Mistral Small 4Mistral
506555506054
Gemma 4 E2BGoogle
526048486252
Kanana 2 30B-A3B ThinkingKakao
407060506052
HyperCLOVAX SEED 1.5BNaver
415139394942
LFM2.5 8B-A1BLiquid AI
425440405444
법무 · 문항 3상속 — 유류분·법정상속분·증여세 (직계존비속 + 배우자)비공개

상속·증여

본문·raw·근거 →
모델
정확성의도 파악신중함한국 맥락짜임새avg
Claude Sonnet 4.6Anthropic
908585929089
Claude Opus 4.8Anthropic
888888908889
GPT-5.5OpenAI
838282848283
MiniMax M3Minimax
889088859288
DeepSeek V4 ProDeepSeek
858282858284
Gemini 3.1 ProGoogle
858282868284
Gemini 3.5 FlashGoogle
828080818081
Mimo V2.5 ProXiaomi
858283858685
Qwen 3.7 MaxAlibaba
888585888687
Kimi K2.6Moonshot
767876748076
GPT-5.4 MiniOpenAI
807880807880
Qwen 3.7 PlusAlibaba
808080808080
Gemma 4 12BGoogle
768480768479
DeepSeek V4 FlashDeepSeek
808078827880
Grok 4.3xAI
807878807879
GLM 5.1Z.ai
727874727874
Step 3.7 FlashStepFun
688578658573
Gemini 3.1 Flash LiteGoogle
807878807679
Gemma 4 26B A4BGoogle
807878807879
Gemma 4 31BGoogle
807878807879
Qwen 3.6 35B A3BAlibaba
506860527458
Qwen 3.6 27BAlibaba
627268627466
EXAONE 4.5 33BLG AI
6080607010072
Nemotron 3 Ultra 550BNVIDIA
458045428854
HyperCLOVAX SEED Think 32BNaver
608060808072
Solar Pro 3Upstage
5080807010072
Qwen 3.5 9BAlibaba
507862487858
Mistral Small 4Mistral
355545355842
Gemma 4 E2BGoogle
374936334738
Kanana 2 30B-A3B ThinkingKakao
406060606055
HyperCLOVAX SEED 1.5BNaver
324432304234
LFM2.5 8B-A1BLiquid AI
203818143822
법무 · 문항 4노동 분쟁 — 부당해고 구제 신청 + 위로금 협상비공개

노동 분쟁 (부당해고)

본문·raw·근거 →
모델
정확성의도 파악신중함한국 맥락짜임새avg
Claude Sonnet 4.6Anthropic
908888909090
Claude Opus 4.8Anthropic
808888888886
GPT-5.5OpenAI
908890909290
MiniMax M3Minimax
829288809585
DeepSeek V4 ProDeepSeek
888585888687
Gemini 3.1 ProGoogle
838582848484
Gemini 3.5 FlashGoogle
868585878886
Mimo V2.5 ProXiaomi
828282828683
Qwen 3.7 MaxAlibaba
868686868886
Kimi K2.6Moonshot
788582788680
GPT-5.4 MiniOpenAI
858585858685
Qwen 3.7 PlusAlibaba
808080808080
Gemma 4 12BGoogle
788684788681
DeepSeek V4 FlashDeepSeek
757875767876
Grok 4.3xAI
828080828081
GLM 5.1Z.ai
748078728276
Step 3.7 FlashStepFun
708580688575
Gemini 3.1 Flash LiteGoogle
808080807880
Gemma 4 26B A4BGoogle
787878788078
Gemma 4 31BGoogle
807878808080
Qwen 3.6 35B A3BAlibaba
657872668070
Qwen 3.6 27BAlibaba
788080788079
EXAONE 4.5 33BLG AI
6080606010068
Nemotron 3 Ultra 550BNVIDIA
508460589064
HyperCLOVAX SEED Think 32BNaver
608060608065
Solar Pro 3Upstage
406040408048
Qwen 3.5 9BAlibaba
358042388048
Mistral Small 4Mistral
305035305536
Gemma 4 E2BGoogle
394937364940
Kanana 2 30B-A3B ThinkingKakao
204040206031
HyperCLOVAX SEED 1.5BNaver
254125203827
LFM2.5 8B-A1BLiquid AI
11351373516
법무 · 문항 5임대차 — 보증금 반환·계약 갱신 (상가건물임대차보호법)비공개

임대차 분쟁

본문·raw·근거 →
모델
정확성의도 파악신중함한국 맥락짜임새avg
Claude Sonnet 4.6Anthropic
888888889088
Claude Opus 4.8Anthropic
888885909089
GPT-5.5OpenAI
908890909290
MiniMax M3Minimax
889288849588
DeepSeek V4 ProDeepSeek
888686888888
Gemini 3.1 ProGoogle
868686878887
Gemini 3.5 FlashGoogle
858585868686
Mimo V2.5 ProXiaomi
858484858685
Qwen 3.7 MaxAlibaba
858585868686
Kimi K2.6Moonshot
868686878887
GPT-5.4 MiniOpenAI
848484848484
Qwen 3.7 PlusAlibaba
8080808010083
Gemma 4 12BGoogle
788682788681
DeepSeek V4 FlashDeepSeek
828280828282
Grok 4.3xAI
848484848484
GLM 5.1Z.ai
868686878887
Step 3.7 FlashStepFun
859085829085
Gemini 3.1 Flash LiteGoogle
808080807880
Gemma 4 26B A4BGoogle
828282828482
Gemma 4 31BGoogle
808080808280
Qwen 3.6 35B A3BAlibaba
748078748076
Qwen 3.6 27BAlibaba
768078748077
EXAONE 4.5 33BLG AI
6080608010075
Nemotron 3 Ultra 550BNVIDIA
758582809081
HyperCLOVAX SEED Think 32BNaver
608060608065
Solar Pro 3Upstage
5080607010069
Qwen 3.5 9BAlibaba
588572658569
Mistral Small 4Mistral
788080778078
Gemma 4 E2BGoogle
505546465850
Kanana 2 30B-A3B ThinkingKakao
202040206029
HyperCLOVAX SEED 1.5BNaver
404838364640
LFM2.5 8B-A1BLiquid AI
324830284834