AI 모델 인텔리전스

기능 · 2026-08-13

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 506개

모델벤더입력 / 1M출력 / 1M컨텍스트제공자
Voxtral Small 24B 2507Mistral$0.002$0.00232K4
Llama 3.2 3B InstructMeta$0.020$0.020131K10
Ling-2.6-flashopenrouter$0.010$0.030262K1
Llama 3.1 8B InstructMeta$0.020$0.030131K18
Mistral Nemo Instruct 2407Mistral$0.020$0.030131K7
Llama 3 8B LunarisMeta$0.040$0.0508K2
Sao10k L3 8B Lunaris novita-ai$0.050$0.0508K1
MythoMax 13Bopenrouter$0.060$0.0608K1
MythoMax 13Bkilo$0.060$0.0604K1
Gemma 4 E4B ITGoogle$0.020$0.100131K3
Nex N2 Mininano-gpt$0.025$0.100262K1
Nex-N2-Miniopenrouter$0.025$0.100262K1
Nex AGI: Nex-N2-Minikilo$0.025$0.100262K1
Qwen3.7 FlashAlibaba (Qwen)$0.030$0.1181M9
Google Gemma 3 12BGoogle$0.050$0.100131K8
Granite 4.1 8Bnano-gpt$0.050$0.100131K1
Granite 4.1 8Bopenrouter$0.050$0.100131K1
IBM: Granite 4.1 8Bkilo$0.050$0.100131K1
Mellum2 12B A2.5Bwandb$0.050$0.100131K1
Granite 4.1 8Bwandb$0.050$0.100131K1
gpt-oss-20bOpenAI$0.030$0.130128K28
GPT OSS 120Bllmgateway$0.032$0.140131K1
Gemma 3n 4BGoogle$0.060$0.12033K4
Qwen3.5 9BAlibaba (Qwen)$0.040$0.150262K22
GPT OSS 20Bllmgateway$0.040$0.150131K1
nova-micro-v1cortecs$0.040$0.159128K1
gpt-oss-120bOpenAI$0.030$0.170128K43
Ministral 3 3B 2512Mistral$0.100$0.100131K3
Reka Edgeopenrouter$0.100$0.10016K1
Ministral 3Bllmgateway$0.100$0.100131K1
Reka Edgekilo$0.100$0.10016K1
GPT OSS 20Bcortecs$0.045$0.167131K1
Mistral-7B-Instruct-v0.3Mistral$0.110$0.11066K5
ministral-3b-2512cortecs$0.111$0.111256K1
Google Gemma 3 27B InstructGoogle$0.080$0.160203K11
Nemotron 3.5 Lightning 30B A3BNVIDIA$0.050$0.2001M3
GPT OSS 20Bdatabricks$0.050$0.200131K1
GPT OSS 20Bneon$0.050$0.200131K1
DeepSeek V4 Flash 0731DeepSeek$0.080$0.1801.05M27
GPT OSS Safeguard 20BOpenAI$0.070$0.200128K6
GPT OSS 20Bfrogbot$0.070$0.200131K1
nvidia-nemotron-3-nano-omniNVIDIA$0.059$0.237300K4
Ministral 3 8B 2512Mistral$0.150$0.150262K3
Ministral 8Bllmgateway$0.150$0.150262K1
Muse Spark 1.2 ContributorMeta$0.100$0.2001.05M3
LFM2.5 2.6Bnano-gpt$0.100$0.200128K1
Reka Flash 3openrouter$0.100$0.20066K1
Reka Flash 3kilo$0.100$0.20066K1
Qwen3.5 FlashAlibaba (Qwen)$0.029$0.2871M7
Tencent Hy3nano-gpt$0.066$0.260262K1
Granite-4.0-H-Smallwatsonx$0.064$0.265131K1
DeepSeek V4 Flash LatestDeepSeek$0.080$0.2521.05M3
ministral-8b-2512cortecs$0.167$0.167256K1
Mistral Small 3.2 24BMistral$0.094$0.250256K3
nova-lite-v1cortecs$0.069$0.275300K1
Nemotron 3.5 Lightning 30B A3BNVIDIA$0.100$0.250262K4
GPT OSS 120Bdatabricks$0.072$0.280131K1
GPT OSS 120Bneon$0.072$0.280131K1
GPT OSS 20Bimpossibl$0.070$0.300131K2
Seed 1.6 Flash (250715)llmgateway$0.070$0.300256K1

전체 506개 중 상위 60개 표시. 추가 필터링은 전체 목록을 이용하세요.

Frequently asked questions

How many AI models support 구조화 출력?

506 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?

Voxtral Small 24B 2507 from Mistral is currently the lowest-priced option, at $0.002 per 1M input tokens and $0.002 per 1M output tokens. The full table above is sorted price-ascending.

Which model with 구조화 출력 has the largest context window?

Pokee-Isaac 28B (nano-gpt) 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-5.2 (74), Kimi K2.6 (62), DeepSeek V4 Pro (55).

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.