구조화된 출력

구조화된 출력 (Structured Output) /v1/messages

LiteLLM을 사용해 /v1/messages 엔드포인트를 통해 Anthropic의 구조화된 출력 기능을 호출할 수 있어요.

지원 프로바이더

사용법

LiteLLM Proxy Server

• Anthropic
• Azure AI (Anthropic)
• Bedrock (Converse)
• Bedrock (Invoke)

1. config.yaml 설정 (Anthropic):

model_list:
  - model_name: claude-sonnet
    litellm_params:
      model: anthropic/claude-sonnet-5
      api_key: os.environ/ANTHROPIC_API_KEY

2. 프록시 시작

litellm --config /path/to/config.yaml

3. 테스트!

curl http://localhost:4000/v1/messages \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer ***" \
  -H "anthropic-version: 2023-06-01" \
  -d '{
    "model": "claude-sonnet",
    "max_tokens": 1024,
    "messages": [
      {
        "role": "user",
        "content": "Extract the key information from this email: John Smith ([email protected]) is interested in our Enterprise plan and wants to schedule a demo for next Tuesday at 2pm."
      }
    ],
    "output_format": {
      "type": "json_schema",
      "schema": {
        "type": "object",
        "properties": {
          "name": {"type": "string"},
          "email": {"type": "string"},
          "plan_interest": {"type": "string"},
          "demo_requested": {"type": "boolean"}
        },
        "required": ["name", "email", "plan_interest", "demo_requested"],
        "additionalProperties": false
      }
    }
  }'

Azure AI (Anthropic):

model_list:
  - model_name: azure-claude-sonnet
    litellm_params:
      model: azure_ai/claude-sonnet-5
      api_key: os.environ/AZURE_AI_API_KEY
      api_base: https://your-endpoint.inference.ai.azure.com

테스트:

curl http://localhost:4000/v1/messages \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer ***" \
  -H "anthropic-version: 2023-06-01" \
  -d '{
    "model": "azure-claude-sonnet",
    "max_tokens": 1024,
    "messages": [
      {
        "role": "user",
        "content": "Extract the key information from this email: John Smith ([email protected]) is interested in our Enterprise plan and wants to schedule a demo for next Tuesday at 2pm."
      }
    ],
    "output_format": {
      "type": "json_schema",
      "schema": {
        "type": "object",
        "properties": {
          "name": {"type": "string"},
          "email": {"type": "string"},
          "plan_interest": {"type": "string"},
          "demo_requested": {"type": "boolean"}
        },
        "required": ["name", "email", "plan_interest", "demo_requested"],
        "additionalProperties": false
      }
    }
  }'

Bedrock (Converse):

model_list:
  - model_name: bedrock-claude-sonnet
    litellm_params:
      model: bedrock/global.anthropic.claude-sonnet-5
      aws_access_key_id: os.environ/AWS_ACCESS_KEY_ID
      aws_secret_access_key: os.environ/AWS_SECRET_ACCESS_KEY
      aws_region_name: us-west-2

테스트:

curl http://localhost:4000/v1/messages \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer ***" \
  -H "anthropic-version: 2023-06-01" \
  -d '{
    "model": "bedrock-claude-sonnet",
    "max_tokens": 1024,
    "messages": [
      {
        "role": "user",
        "content": "Extract the key information from this email: John Smith ([email protected]) is interested in our Enterprise plan and wants to schedule a demo for next Tuesday at 2pm."
      }
    ],
    "output_format": {
      "type": "json_schema",
      "schema": {
        "type": "object",
        "properties": {
          "name": {"type": "string"},
          "email": {"type": "string"},
          "plan_interest": {"type": "string"},
          "demo_requested": {"type": "boolean"}
        },
        "required": ["name", "email", "plan_interest", "demo_requested"],
        "additionalProperties": false
      }
    }
  }'

Bedrock (Invoke):

model_list:
  - model_name: bedrock-claude-invoke
    litellm_params:
      model: bedrock/invoke/global.anthropic.claude-sonnet-5
      aws_access_key_id: os.environ/AWS_ACCESS_KEY_ID
      aws_secret_access_key: os.environ/AWS_SECRET_ACCESS_KEY
      aws_region_name: us-west-2

테스트:

curl http://localhost:4000/v1/messages \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer ***" \
  -H "anthropic-version: 2023-06-01" \
  -d '{
    "model": "bedrock-claude-invoke",
    "max_tokens": 1024,
    "messages": [
      {
        "role": "user",
        "content": "Extract the key information from this email: John Smith ([email protected]) is interested in our Enterprise plan and wants to schedule a demo for next Tuesday at 2pm."
      }
    ],
    "output_format": {
      "type": "json_schema",
      "schema": {
        "type": "object",
        "properties": {
          "name": {"type": "string"},
          "email": {"type": "string"},
          "plan_interest": {"type": "string"},
          "demo_requested": {"type": "boolean"}
        },
        "required": ["name", "email", "plan_interest", "demo_requested"],
        "additionalProperties": false
      }
    }
  }'

예시 응답

{
  "id": "msg_01XFDUDYJgAACzvnptvVoYEL",
  "type": "message",
  "role": "assistant",
  "content": [
    {
      "type": "text",
      "text": "{\"name\":\"John Smith\",\"email\":\"[email protected]\",\"plan_interest\":\"Enterprise\",\"demo_requested\":true}"
    }
  ],
  "model": "claude-sonnet-5",
  "stop_reason": "end_turn",
  "stop_sequence": null,
  "usage": {
    "input_tokens": 75,
    "output_tokens": 28
  }
}

요청 형식

output_format

output_format 파라미터는 구조화된 출력 형식을 지정해요.

{
  "output_format": {
    "type": "json_schema",
    "schema": {
      "type": "object",
      "properties": {
        "field_name": {"type": "string"},
        "another_field": {"type": "integer"}
      },
      "required": ["field_name", "another_field"],
      "additionalProperties": false
    }
  }
}

필드

• type (string): 반드시 "json_schema" 여야 합니다
• schema (object): 예상 출력 구조를 정의하는 JSON Schema 객체
  • type (string): 루트 타입, 보통 "object"
  • properties (object): 필드와 그 타입 정의
  • required (array): 필수 필드 이름 목록
  • additionalProperties (boolean): 엄격한 스키마 준수를 강제하려면 false로 설정

출처: 문서

더 알아보기 (Learn more)

• /v1/messages
• /v1/messages Parameter Mapping