AI 모델 인텔리전스

기능 · 2026-11-12

구조화 출력을 지원하는 AI 모델

JSON mode 또는 스키마 준수 출력을 지원하는 AI 모델 비교 — 추출 및 분류 파이프라인에 적합.

이게 뭔가요?

  • 구조화 출력(JSON mode 또는 response_format=json_schema)은 모델이 제공한 스키마에 맞는 JSON 문서만 생성하도록 강제합니다.
  • 프롬프트에서 'JSON으로 답해줘'라고 요청하는 것과 달리, 디코딩 시점에 적용되어 모델이 유효하지 않은 JSON을 출력할 수 없습니다.

왜 중요한가

  • 파싱 오류와 '네, JSON은 다음과 같습니다…' 같은 접두사를 줄여줍니다.
  • LLM 출력을 타입 시스템에 연결하는 파이프라인에 필수: 추출, 분류, 구조화 요약 등.

이 기능을 지원하는 모델 914개

모델벤더입력 / 1M출력 / 1M컨텍스트제공자
Qwen3.8 27BAlibaba (Qwen)$0.001$0.0011.05M51
Llama 3.2 3B InstructMeta$0.020$0.020131K7
Gemma 4 26B A4B ITGoogle$0.010$0.031262K23
Llama-3.1-8B-InstructMeta$0.020$0.030131K14
Mistral NemoMistral$0.029$0.030131K6
inclusionAI: Ling 3.0 Tinykilo$0.015$0.045262K1
Ling 3.0 Tinyopenrouter$0.015$0.045262K1
Llama 3 8B LunarisMeta$0.040$0.0508K3
Qwen3.5 9BAlibaba (Qwen)$0.040$0.060262K21
gpt-oss-20bOpenAI$0.018$0.090131K32
Llama-3.2-11B-Vision-Instruct (Deep Infra)Meta$0.055$0.055131K4
Qwen3.5 4BAlibaba (Qwen)$0.040$0.070262K5
GLM Flash LatestZ.AI / Zhipu$0.040$0.0781.05M4
Nex AGI: Nex-N2.5-Minikilo$0.025$0.100262K1
Nex N2.5 Mininano-gpt$0.025$0.100262K1
Nex-N2.5-Miniopenrouter$0.025$0.100262K1
Mistral Small 3Mistral$0.050$0.08033K6
DeepSeek V4 Flash 0731DeepSeek$0.045$0.0901.05M54
Nemotron 3.5 Lightning 30B A3BNVIDIA$0.025$0.120262K5
DeepSeek-V4.1-FlashDeepSeek$0.030$0.1201M72
DeepSeek V4.1 Flash 1017DeepSeek$0.030$0.1201.05M60
Gemma 3 12B ITGoogle$0.050$0.100131K13
Qwen3.7 FlashAlibaba (Qwen)$0.030$0.1301M13
Nova Micro (US)edenai$0.035$0.140128K1
Nova Microedenai$0.035$0.140128K1
gpt-oss-120bOpenAI$0.030$0.150131K51
Inference.net: Schematron V2 Turbokilo$0.030$0.150128K1
Schematron V2 Turbonano-gpt$0.030$0.150128K1
Schematron V2 Turbovercel$0.030$0.150128K1
Schematron V2 Turboopenrouter$0.030$0.150128K1
Command R7BCohere$0.037$0.150128K6
Command R7B ArabicCohere$0.037$0.150128K1
MythoMax 13Bkilo$0.080$0.1104K1
Mercury 2.5 Previewinception$0.040$0.150260K1
MythoMax L2 13Bopenrouter$0.080$0.1108K1
Schematron V2 Turboaimlapi$0.032$0.159128K1
Nova Micro 1.0cortecs$0.040$0.159128K1
Ministral 3 3BMistral$0.100$0.100131K8
Nemotron 3.5 Lightning 30B A3BNVIDIA$0.050$0.150262K7
Reka Edge 2603 (reka-ai)llmgateway-providers$0.100$0.10016K1
Reka Edgekilo$0.100$0.10016K1
Ministral 3B (latest)Mistral$0.100$0.100131K1
Reka Edge 2603llmgateway$0.100$0.10016K1
Ministral 3Bllmgateway$0.100$0.100131K1
Reka Edgeopenrouter$0.100$0.10016K1
IBM: Granite 4.2 8Bkilo$0.060$0.150131K1
Granite 4.2 8Bnano-gpt$0.060$0.150131K1
Reka Edge 2603nano-gpt$0.105$0.10516K1
Granite 4.2 8Bwandb$0.060$0.150131K1
Granite 4.2 8Bopenrouter$0.060$0.150131K1
PrismML: Ternary Bonsai 2 9Bkilo$0.080$0.130262K1
Ternary Bonsai 2 9Bopenrouter$0.080$0.130262K1
Reka Edgeaimlapi$0.106$0.10616K1
GPT OSS 20Bcortecs$0.045$0.167131K1
GPT OSS 20B (OVHcloud)edenai$0.050$0.180131K3
GPT OSS 20Bllmgateway$0.040$0.190131K1
GPT OSS 120Bkenari$0.036$0.200131K1
Gemma 3 27B ITGoogle$0.080$0.160131K19
Ling 3.0 Flash VL (DeepInfra)llmgateway-providers$0.060$0.180131K1
inclusionAI: Ling 3.0 Flash VLkilo$0.060$0.180131K1

전체 914개 중 상위 60개 표시. 추가 필터링은 전체 목록을 이용하세요.

Frequently asked questions

How many AI models support 구조화 출력?

914 canonical models in our database currently support 구조화 출력. The list is regenerated on every data refresh, so it always reflects the latest releases tracked in our catalogue.

What is the cheapest model with 구조화 출력?

Qwen3.8 27B from Alibaba (Qwen) is currently the lowest-priced option, at $0.001 per 1M input tokens and $0.001 per 1M output tokens. The full table above is sorted price-ascending.

Which model with 구조화 출력 has the largest context window?

Qwen Long (Alibaba (Qwen)) leads on context at 10M tokens. This may matter if you also need long-document understanding alongside 구조화 출력.

Which models are available on the most providers?

Production-readiness usually correlates with how many independent providers host the same weights. The top three by provider count are: GLM-5.3-Flash (92), GLM-5.3 (92), GLM-5.2 (90).

How is 구조화 출력 different from a regular LLM?

Structured output (a.k.a. JSON mode / response_format=json_schema) constrains the model at decode time so it cannot emit invalid JSON. This is stricter than just prompting 'reply in JSON' and removes a whole class of parsing errors.

How often is this list updated?

Daily. Our data pipeline syncs once a day, regenerates the canonical model list, and rebuilds these pages so newly released models appear within 24 hours.

마지막 업데이트:

Prices in USD per 1M tokens. Unknown means the provider does not publish per-token pricing.

Pricing and capabilities are refreshed daily and reconciled against each provider's official documentation. Always verify critical production decisions with the provider directly.