使用 Claude 構建

使用 Messages API

有效使用 Messages API 的實用模式和範例

本指南涵蓋使用 Messages API 的常見模式，包括基本請求、多輪對話、預填充技術和視覺功能。如需完整的 API 規格，請參閱 Messages API 參考。

基本請求和回應

#!/bin/sh
curl https://api.anthropic.com/v1/messages \
     --header "x-api-key: $ANTHROPIC_API_KEY" \
     --header "anthropic-version: 2023-06-01" \
     --header "content-type: application/json" \
     --data \
'{
    "model": "claude-sonnet-4-5",
    "max_tokens": 1024,
    "messages": [
        {"role": "user", "content": "Hello, Claude"}
    ]
}'

JSON

{
  "id": "msg_01XFDUDYJgAACzvnptvVoYEL",
  "type": "message",
  "role": "assistant",
  "content": [
    {
      "type": "text",
      "text": "Hello!"
    }
  ],
  "model": "claude-sonnet-4-5",
  "stop_reason": "end_turn",
  "stop_sequence": null,
  "usage": {
    "input_tokens": 12,
    "output_tokens": 6
  }
}

多輪對話

Messages API 是無狀態的，這意味著您始終將完整的對話歷史發送到 API。您可以使用此模式隨著時間推移建立對話。早期的對話輪次不一定需要實際來自 Claude — 您可以使用合成的 assistant 訊息。

JSON

{
    "id": "msg_018gCsTGsXkYJVqYPxTgDHBU",
    "type": "message",
    "role": "assistant",
    "content": [
        {
            "type": "text",
            "text": "Sure, I'd be happy to provide..."
        }
    ],
    "stop_reason": "end_turn",
    "stop_sequence": null,
    "usage": {
      "input_tokens": 30,
      "output_tokens": 309
    }
}

將話語放入 Claude 的嘴裡

您可以在輸入訊息列表的最後位置預填充 Claude 回應的一部分。這可用於塑造 Claude 的回應。下面的範例使用 "max_tokens": 1 從 Claude 獲得單一選擇題答案。

JSON

{
  "id": "msg_01Q8Faay6S7QPTvEUUQARt7h",
  "type": "message",
  "role": "assistant",
  "content": [
    {
      "type": "text",
      "text": "C"
    }
  ],
  "model": "claude-sonnet-4-5",
  "stop_reason": "max_tokens",
  "stop_sequence": null,
  "usage": {
    "input_tokens": 42,
    "output_tokens": 1
  }
}

如需更多關於預填充技術的資訊，請參閱我們的預填充指南。

視覺

Claude 可以在請求中讀取文字和圖像。我們支援圖像的 base64 和 url 來源類型，以及 image/jpeg、image/png、image/gif 和 image/webp 媒體類型。如需更多詳細資訊，請參閱我們的視覺指南。

JSON

{
  "id": "msg_01EcyWo6m4hyW8KHs2y2pei5",
  "type": "message",
  "role": "assistant",
  "content": [
    {
      "type": "text",
      "text": "This image shows an ant, specifically a close-up view of an ant. The ant is shown in detail, with its distinct head, antennae, and legs clearly visible. The image is focused on capturing the intricate details and features of the ant, likely taken with a macro lens to get an extreme close-up perspective."
    }
  ],
  "model": "claude-sonnet-4-5",
  "stop_reason": "end_turn",
  "stop_sequence": null,
  "usage": {
    "input_tokens": 1551,
    "output_tokens": 71
  }
}

工具使用和電腦使用

請參閱我們的指南以了解如何在 Messages API 中使用工具的範例。請參閱我們的電腦使用指南以了解如何使用 Messages API 控制桌面電腦環境的範例。如需保證的 JSON 輸出，請參閱結構化輸出。

基本請求和回應

#!/bin/sh
curl https://api.anthropic.com/v1/messages \
     --header "x-api-key: $ANTHROPIC_API_KEY" \
     --header "anthropic-version: 2023-06-01" \
     --header "content-type: application/json" \
     --data \
'{
    "model": "claude-sonnet-4-5",
    "max_tokens": 1024,
    "messages": [
        {"role": "user", "content": "Hello, Claude"}
    ]
}'

JSON

{
  "id": "msg_01XFDUDYJgAACzvnptvVoYEL",
  "type": "message",
  "role": "assistant",
  "content": [
    {
      "type": "text",
      "text": "Hello!"
    }
  ],
  "model": "claude-sonnet-4-5",
  "stop_reason": "end_turn",
  "stop_sequence": null,
  "usage": {
    "input_tokens": 12,
    "output_tokens": 6
  }
}

多輪對話

JSON

{
    "id": "msg_018gCsTGsXkYJVqYPxTgDHBU",
    "type": "message",
    "role": "assistant",
    "content": [
        {
            "type": "text",
            "text": "Sure, I'd be happy to provide..."
        }
    ],
    "stop_reason": "end_turn",
    "stop_sequence": null,
    "usage": {
      "input_tokens": 30,
      "output_tokens": 309
    }
}

將話語放入 Claude 的嘴裡

JSON

{
  "id": "msg_01Q8Faay6S7QPTvEUUQARt7h",
  "type": "message",
  "role": "assistant",
  "content": [
    {
      "type": "text",
      "text": "C"
    }
  ],
  "model": "claude-sonnet-4-5",
  "stop_reason": "max_tokens",
  "stop_sequence": null,
  "usage": {
    "input_tokens": 42,
    "output_tokens": 1
  }
}

如需更多關於預填充技術的資訊，請參閱我們的預填充指南。

視覺

JSON

{
  "id": "msg_01EcyWo6m4hyW8KHs2y2pei5",
  "type": "message",
  "role": "assistant",
  "content": [
    {
      "type": "text",
      "text": "This image shows an ant, specifically a close-up view of an ant. The ant is shown in detail, with its distinct head, antennae, and legs clearly visible. The image is focused on capturing the intricate details and features of the ant, likely taken with a macro lens to get an extreme close-up perspective."
    }
  ],
  "model": "claude-sonnet-4-5",
  "stop_reason": "end_turn",
  "stop_sequence": null,
  "usage": {
    "input_tokens": 1551,
    "output_tokens": 71
  }
}