聊天接口 - API2D

这里只列出部分常用参数，完整参数说明参考 Claude 官方文档：https://docs.anthropic.com/claude/reference/messages_post

模型定价均为官方定价的 1.5 倍。

支持用同样的方式设置和使用 API2D 提供的【搜索】、【知识库】、【联网】功能。

已经支持 claude-3-7-sonnet-20250219，claude-3-7-sonnet-latest，claude-3-5-haiku-20241022，claude-3-5-haiku-latest，claude-3-5-sonnet-20241022，claude-3-5-sonnet-latest，claude-3-opus-latest。如果使用 OpenAI 兼容模式（API 地址设置为 https://oa.api2d.net/claude ），所有 GPT-3.5 都会被映射为 claude-3-5-sonnet-latest，所有 GPT-4 都会被映射为 claude-3-opus-latest。

请求参数

Header 参数

x-api-key

string

必需

x-api-key和Authorization作用相同，选一个使用即可。x-api-key主要是为了兼容Claude的参数格式。

示例值:

fk186009-...

Content-Type

string

必需

默认值:

application/json

Authorization

string

必需

Bearer 你的forwardKey/customKey/API Token

默认值:

Bearer fk....

Body 参数application/json

model

enum<string>

必需

枚举值:

claude-3-opus-20240229claude-3-sonnet-20240229claude-3-haiku-20240307claude-2.1claude-2.0claude-instant-1.2claude-3-7-sonnet-20250219claude-3-7-sonnet-latestclaude-3-5-haiku-20241022claude-3-5-haiku-latestclaude-3-5-sonnet-20241022claude-3-5-sonnet-latest

messages

array [object {2}]

必需

<= 1 items

role

string

必需

content

string

必需

stream

boolean

是否启用流式返回

可选

默认值:

false

max_tokens

integer

可选

如果不指定，每次请求会预先冻结模型支持的最大生成 Token（目前是 4096），在请求完成后按 usage 字段多退少补。如果账户余额不够冻结就会报错，此时可以指定一个比较小的 max_tokens，这样就会按 max_tokens 来冻结

示例

{
  "model": "claude-3-opus-20240229",
  "messages": [
    {
      "role": "user",
      "content": "讲个笑话"
    }
  ],
  "stream": true,
  "max_tokens": 100
}

示例代码

Shell

JavaScript

Java

Swift

PHP

Python

HTTP

Objective-C

Ruby

OCaml

Dart

返回响应

🟢200OK

application/json

Body

object {0}

示例

event: message_start
data: {"type":"message_start","message":{"id":"msg_01GbTJMWgepqJMBQedFmZybG","type":"message","role":"assistant","content":[],"model":"claude-instant-1.2","stop_reason":null,"stop_sequence":null,"usage":{"input_tokens":16,"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":"今"}}

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

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

event: content_block_delta
data: {"type":"content_block_delta","index":0,"delta":{"type":"text_delta","text":"2022"}}

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

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":11}}

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