在创建消息时，您可以设置 "stream": true 来使用 server-sent events (SSE) 以增量方式流式传输响应。

​
使用 SDK 进行流式传输

我们的 Python 和 Typescript SDK 提供了多种流式传输方式。Python SDK 允许同步和异步流。有关详细信息，请参阅每个 SDK 中的文档。

import anthropic

client = anthropic.Anthropic()

with client.messages.stream(
    max_tokens=1024,
    messages=[{"role": "user", "content": "Hello"}],
    model="claude-3-opus-20240229",
) as stream:
  for text in stream.text_stream:
      print(text, end="", flush=True)

​
事件类型

每个服务器发送的事件都包括一个命名的事件类型和相关的 JSON 数据。每个事件都将使用 SSE 事件名称（例如 event: message_stop），并在其数据中包含匹配的事件 type。

每个流使用以下事件流程：

  1. message_start：包含一个 Message 对象，其中 content 为空。
  2. 一系列内容块，每个内容块都有一个 content_block_start、一个或多个 content_block_delta 事件和一个 content_block_stop 事件。每个内容块都有一个 index，对应于其在最终 Message content 数组中的索引。
  3. 一个或多个 message_delta 事件，表示对最终 Message 对象的顶层更改。
  4. 最后一个 message_stop 事件。

​
Ping 事件

事件流还可能包括任意数量的 ping 事件。

​
错误事件

我们偶尔可能会在事件流中发送错误。例如，在使用量高的时期，您可能会收到一个 overloaded_error，这通常对应于非流式上下文中的 HTTP 529：

错误示例
event: error
data: {"type": "error", "error": {"type": "overloaded_error", "message": "Overloaded"}}

​
其他事件

根据我们的版本控制策略，我们可能会添加新的事件类型，您的代码应该优雅地处理未知的事件类型。

​
Delta 类型

每个 content_block_delta 事件都包含一个 delta，其类型更新给定 index 处的 content 块。

​
文本 delta

text 内容块 delta 如下所示：

文本 delta
event: content_block_delta
data: {"type": "content_block_delta","index": 0,"delta": {"type": "text_delta", "text": "ello frien"}}

​
输入 JSON delta

tool_use 内容块的 delta 对应于块的 input 字段的更新。为了支持最大粒度，delta 是 部分 JSON 字符串，而最终的 tool_use.input 始终是一个 对象。

您可以累积字符串 delta，并在收到 content_block_stop 事件时使用像 Pydantic 这样的库进行部分 JSON 解析，或者使用我们的 SDK，它们提供了访问解析后的增量值的帮助程序。

tool_use 内容块 delta 如下所示：

输入 JSON delta
event: content_block_delta
data: {"type": "content_block_delta","index": 1,"delta": {"type": "input_json_delta","partial_json": "{\"location\": \"San Fra"}}}

注意：我们当前的模型只支持一次从 input 发出一个完整的键和值属性。因此，在使用工具时，流事件之间可能会有延迟，而模型正在工作。一旦累积了一个 input 键和值，我们就会将它们作为多个 content_block_delta 事件发出，并使用分块的部分 json，以便该格式可以在将来的模型中自动支持更细粒度。

​
原始 HTTP 流响应

我们强烈建议在使用流模式时使用我们的客户端 SDK。但是，如果您正在构建直接的 API 集成，则需要自己处理这些事件。

流响应包括：

  1. 一个 message_start 事件
  2. 可能有多个内容块，每个内容块包含： a. 一个 content_block_start 事件 b. 可能有多个 content_block_delta 事件 c. 一个 content_block_stop 事件
  3. 一个 message_delta 事件
  4. 一个 message_stop 事件

响应中可能会分散 ping 事件。有关格式的更多详细信息，请参阅事件类型。

​
基本流请求

请求
curl https://api.anthropic.com/v1/messages \
     --header "anthropic-version: 2023-06-01" \
     --header "content-type: application/json" \
     --header "x-api-key: $ANTHROPIC_API_KEY" \
     --data \
'{
  "model": "claude-3-opus-20240229",
  "messages": [{"role": "user", "content": "Hello"}],
  "max_tokens": 256,
  "stream": true
}'
响应
event: message_start
data: {"type": "message_start", "message": {"id": "msg_1nZdL29xx5MUA1yADyHTEsnR8uuvGzszyY", "type": "message", "role": "assistant", "content": [], "model": "claude-3-opus-20240229", "stop_reason": null, "stop_sequence": null, "usage": {"input_tokens": 25, "output_tokens": 1}}}

event: content_block_start
data: {"type": "content_block_start", "index": 0, "content_block": {"type": "text", "text": ""}}

event: ping
data: {"type": "ping"}

event: content_block_delta
data: {"type": "content_block_delta", "index": 0, "delta": {"type": "text_delta", "text": "Hello"}}

event: content_block_delta
data: {"type": "content_block_delta", "index": 0, "delta": {"type": "text_delta", "text": "!"}}

event: content_block_stop
data: {"type": "content_block_stop", "index": 0}

event: message_delta
data: {"type": "message_delta", "delta": {"stop_reason": "end_turn", "stop_sequence":null}, "usage": {"output_tokens": 15}}

event: message_stop
data: {"type": "message_stop"}

​
使用工具的流请求

在此请求中，我们要求 Claude 使用工具告诉我们天气。

请求
  curl https://api.anthropic.com/v1/messages \
    -H "content-type: application/json" \
    -H "x-api-key: $ANTHROPIC_API_KEY" \
    -H "anthropic-version: 2023-06-01" \
    -d '{
      "model": "claude-3-opus-20240229",
      "max_tokens": 1024,
      "tools": [
        {
          "name": "get_weather",
          "description": "获取给定位置的当前天气",
          "input_schema": {
            "type": "object",
            "properties": {
              "location": {
                "type": "string",
                "description": "城市和州，例如 San Francisco, CA"
              }
            },
            "required": ["location"]
          }
        }
      ],
      "tool_choice": {"type": "any"},
      "messages": [
        {
          "role": "user",
          "content": "旧金山的天气如何？"
        }
      ],
      "stream": true
    }'
响应
event: message_start
data: {"type":"message_start","message":{"id":"msg_014p7gG3wDgGV9EUtLvnow3U","type":"message","role":"assistant","model":"claude-3-haiku-20240307","stop_sequence":null,"usage":{"input_tokens":472,"output_tokens":2},"content":[],"stop_reason":null}}

event: content_block_start
data: {"type":"content_block_start","index":0,"content_block":{"type":"text","text":""}}

event: ping
data: {"type": "ping"}

event: content_block_delta
data: {"type":"content_block_delta","index":0,"delta":{"type":"text_delta","text":"好的"}}

event: content_block_delta
data: {"type":"content_block_delta","index":0,"delta":{"type":"text_delta","text":"，"}}

event: content_block_delta
data: {"type":"content_block_delta","index":0,"delta":{"type":"text_delta","text":"让我们"}}

event: content_block_delta
data: {"type":"content_block_delta","index":0,"delta":{"type":"text_delta","text":"查看"}}

event: content_block_delta
data: {"type":"content_block_delta","index":0,"delta":{"type":"text_delta","text":"一下"}}

event: content_block_delta
data: {"type":"content_block_delta","index":0,"delta":{"type":"text_delta","text":"San"}}

event: content_block_delta
data: {"type":"content_block_delta","index":0,"delta":{"type":"text_delta","text":" Francisco"}}

event: content_block_delta
data: {"type":"content_block_delta","index":0,"delta":{"type":"text_delta","text":"，"}}

event: content_block_delta
data: {"type":"content_block_delta","index":0,"delta":{"type":"text_delta","text":"CA"}}

event: content_block_delta
data: {"type":"content_block_delta","index":0,"delta":{"type":"text_delta","text":"的天气"}}

event: content_block_delta
data: {"type":"content_block_delta","index":0,"delta":{"type":"text_delta","text":"："}}

event: content_block_stop
data: {"type":"content_block_stop","index":0}

event: content_block_start
data: {"type":"content_block_start","index":1,"content_block":{"type":"tool_use","id":"toolu_01T1x1fJ34qAmk2tNTrN7Up6","name":"get_weather","input":{}}}

event: content_block_delta
data: {"type":"content_block_delta","index":1,"delta":{"type":"input_json_delta","partial_json":""}}

event: content_block_delta
data: {"type":"content_block_delta","index":1,"delta":{"type":"input_json_delta","partial_json":"{\"location\":"}}

event: content_block_delta
data: {"type":"content_block_delta","index":1,"delta":{"type":"input_json_delta","partial_json":" \"San"}}

event: content_block_delta
data: {"type":"content_block_delta","index":1,"delta":{"type":"input_json_delta","partial_json":" Francisc"}}

event: content_block_delta
data: {"type":"content_block_delta","index":1,"delta":{"type":"input_json_delta","partial_json":"o,"}}

event: content_block_delta
data: {"type":"content_block_delta","index":1,"delta":{"type":"input_json_delta","partial_json":" CA\""}}

event: content_block_delta
data: {"type":"content_block_delta","index":1,"delta":{"type":"input_json_delta","partial_json":", "}}

event: content_block_delta
data: {"type":"content_block_delta","index":1,"delta":{"type":"input_json_delta","partial_json":"\"unit\": \"fah"}}

event: content_block_delta
data: {"type":"content_block_delta","index":1,"delta":{"type":"input_json_delta","partial_json":"renheit\"}"}}

event: content_block_stop
data: {"type":"content_block_stop","index":1}

event: message_delta
data: {"type":"message_delta","delta":{"stop_reason":"tool_use","stop_sequence":null},"usage":{"output_tokens":89}}

event: message_stop
data: {"type":"message_stop"}
Create a Message从 Text Completions 迁移