OpenAI 格式 - HaiToken Doc

POST

chat

completions

创建聊天补全

curl --request POST \
  --url https://api.example.com/v1/chat/completions \
  --header 'Authorization: <authorization>' \
  --header 'Content-Type: application/json' \
  --data '
{
  "model": "<string>",
  "messages": [
    {
      "tool_calls": [
        {
          "id": "<string>",
          "index": 123,
          "type": "<string>",
          "function": {
            "name": "<string>",
            "arguments": "<string>"
          }
        }
      ],
      "tool_call_id": "<string>",
      "reasoning_content": "<string>",
      "content": {},
      "name": "<string>",
      "refusal": "<string>",
      "annotations": [
        {}
      ]
    }
  ],
  "stream_options": {
    "include_usage": true
  },
  "top_p": 123,
  "max_tokens": 123,
  "max_completion_tokens": 123,
  "frequency_penalty": 123,
  "presence_penalty": 123,
  "logit_bias": {
    "key": 123
  },
  "top_logprobs": 123,
  "response_format": {
    "json_schema": {
      "name": "<string>",
      "description": "<string>",
      "schema": {},
      "strict": true
    },
    "type": "<string>"
  },
  "prompt_cache_key": "<string>",
  "prompt_cache_retention": "<string>",
  "safety_identifier": "<string>",
  "tool_choice": {},
  "parallel_tool_calls": true,
  "service_tier": "<string>",
  "reasoning_effort": "<string>",
  "web_search_options": {},
  "function_call": {},
  "stream": true,
  "temperature": 123,
  "stop": {},
  "n": 123,
  "logprobs": true,
  "seed": 123,
  "user": "<string>",
  "tools": [
    {
      "type": "<string>",
      "function": {
        "name": "<string>",
        "description": "<string>",
        "parameters": {},
        "strict": true
      }
    }
  ],
  "modalities": [
    "<string>"
  ],
  "audio": {},
  "metadata": {
    "key": "<string>"
  },
  "prediction": {},
  "verbosity": "<string>",
  "functions": [
    {}
  ]
}
'

{
  "system_fingerprint": "<string>",
  "service_tier": "<string>",
  "id": "<string>",
  "object": "<string>",
  "created": 123,
  "model": "<string>",
  "choices": [
    {
      "finish_reason": "<string>",
      "index": 123,
      "message": {
        "tool_calls": [
          {
            "id": "<string>",
            "index": 123,
            "type": "<string>",
            "function": {
              "name": "<string>",
              "arguments": "<string>"
            }
          }
        ],
        "tool_call_id": "<string>",
        "reasoning_content": "<string>",
        "content": {},
        "name": "<string>",
        "refusal": "<string>",
        "annotations": [
          {}
        ]
      },
      "logprobs": {
        "content": [
          {
            "top_logprobs": [
              {
                "token": "<string>",
                "logprob": 123,
                "bytes": [
                  123
                ]
              }
            ],
            "token": "<string>",
            "logprob": 123,
            "bytes": [
              123
            ]
          }
        ],
        "refusal": [
          {
            "top_logprobs": [
              {
                "token": "<string>",
                "logprob": 123,
                "bytes": [
                  123
                ]
              }
            ],
            "token": "<string>",
            "logprob": 123,
            "bytes": [
              123
            ]
          }
        ]
      }
    }
  ],
  "usage": {
    "prompt_tokens": 123,
    "completion_tokens": 123,
    "total_tokens": 123,
    "prompt_tokens_details": {
      "cached_tokens": 123,
      "audio_tokens": 123
    },
    "completion_tokens_details": {
      "reasoning_tokens": 123,
      "audio_tokens": 123,
      "accepted_prediction_tokens": 123,
      "rejected_prediction_tokens": 123
    },
    "ephemeral_5m_input_tokens": 123,
    "ephemeral_1h_input_tokens": 123,
    "web_search_requests": 123
  }
}

此接口兼容 OpenAI Chat Completions 协议，用于创建聊天补全请求。

功能特性

支持流式（SSE）和非流式两种响应模式
支持多模态输入（文本、图片等）
支持函数调用与工具调用（Function/Tool Calling）
支持结构化输出（JSON Schema）
支持推理模式配置（reasoning_effort）
支持网络搜索选项

使用场景

适用于需要与 OpenAI 兼容模型进行对话交互的场景，包括：

单轮或多轮对话
函数调用与工具编排
结构化 JSON 输出
流式实时响应

认证方式

在请求头中携带 Authorization 字段，格式为 Bearer YOUR_API_KEY。

请勿在客户端代码中暴露 API Key，建议通过服务端代理转发请求。

快速示例

import requests

url = "https://api.haitoken.ai/v1/chat/completions"
headers = {
    "Authorization": "Bearer YOUR_API_KEY",
    "Content-Type": "application/json"
}
data = {
    "model": "gpt-4o",
    "messages": [
        {"role": "user", "content": "你好，请介绍一下自己"}
    ]
}

response = requests.post(url, json=data, headers=headers)
print(response.json())

请求头

Authorization

string

必填

示例:

"Bearer YOUR_API_KEY"

Content-Type

string

示例:

"application/json"

请求体

application/json

model

string

必填

模型 ID，如 gpt-4o、deepseek-chat 等

messages

object[]

必填

消息列表

Show child attributes

stream_options

object

流式响应选项（仅在 stream=true 时使用）

Show child attributes

top_p

number

核采样，0-1 之间

max_tokens

integer

生成的最大 token 数量（已弃用，建议使用 max_completion_tokens）

max_completion_tokens

integer

生成的最大 token 数量（包括可见输出 token 和推理 token）

frequency_penalty

number

频率惩罚，-2.0 到 2.0

presence_penalty

number

存在惩罚，-2.0 到 2.0

logit_bias

object

修改指定 token 出现概率的映射

Show child attributes

top_logprobs

integer

每个 token 位置返回的最可能 token 的最大数量（0-20）

response_format

object

响应格式，支持 "text"、"json_object"、"json_schema"

Show child attributes

prompt_cache_key

string

稳定标识符，用于缓存优化

prompt_cache_retention

string

缓存保留策略："in_memory" 或 "24h"

safety_identifier

string

安全标识符，用于检测违规用户

tool_choice

object

控制模型调用哪个工具，可以是字符串（"none"/"auto"/"required"）或对象

parallel_tool_calls

boolean

是否启用并行工具调用

service_tier

string

服务层级："auto"、"default"、"flex"、"scale"、"priority"

reasoning_effort

string

推理努力程度："none"、"minimal"、"low"、"medium"、"high"、"xhigh"

web_search_options

object

网络搜索选项

function_call

object

已弃用：函数调用控制

stream

boolean

是否启用流式输出

temperature

number

采样温度，0-2 之间，值越高输出越随机

stop

object

停止序列

integer

每个输入消息生成的聊天补全选择数量

logprobs

boolean

是否返回输出 token 的对数概率

seed

integer<int64>

用于确定性采样的种子值

user

string

用户标识（已弃用，建议使用 prompt_cache_key）

tools

object[]

模型可调用的工具列表

Show child attributes

modalities

string[]

输出类型列表：["text"] 或 ["text", "audio"]

audio

object

音频输出参数

metadata

object

元数据，最多16个键值对

Show child attributes

prediction

object

预测输出内容

verbosity

string

响应详细程度："low"、"medium"、"high"

functions

object[]

已弃用：函数定义列表

响应

200 - application/json

OpenAI Chat Completions API 响应对象

system_fingerprint

string

系统指纹，表示后端配置

service_tier

string

实际使用的服务层级

string

聊天补全的唯一标识

object

string

对象类型，始终为 "chat.completion"

created

integer<int64>

创建时间（Unix 时间戳）

model

string

使用的模型 ID

choices

object[]

聊天补全选择列表

Show child attributes

usage

object

使用统计信息

Show child attributes

Claude 格式创建 Claude 消息，兼容 Anthropic Messages 协议，支持流式和非流式响应

创建聊天补全

curl --request POST \
  --url https://api.example.com/v1/chat/completions \
  --header 'Authorization: <authorization>' \
  --header 'Content-Type: application/json' \
  --data '
{
  "model": "<string>",
  "messages": [
    {
      "tool_calls": [
        {
          "id": "<string>",
          "index": 123,
          "type": "<string>",
          "function": {
            "name": "<string>",
            "arguments": "<string>"
          }
        }
      ],
      "tool_call_id": "<string>",
      "reasoning_content": "<string>",
      "content": {},
      "name": "<string>",
      "refusal": "<string>",
      "annotations": [
        {}
      ]
    }
  ],
  "stream_options": {
    "include_usage": true
  },
  "top_p": 123,
  "max_tokens": 123,
  "max_completion_tokens": 123,
  "frequency_penalty": 123,
  "presence_penalty": 123,
  "logit_bias": {
    "key": 123
  },
  "top_logprobs": 123,
  "response_format": {
    "json_schema": {
      "name": "<string>",
      "description": "<string>",
      "schema": {},
      "strict": true
    },
    "type": "<string>"
  },
  "prompt_cache_key": "<string>",
  "prompt_cache_retention": "<string>",
  "safety_identifier": "<string>",
  "tool_choice": {},
  "parallel_tool_calls": true,
  "service_tier": "<string>",
  "reasoning_effort": "<string>",
  "web_search_options": {},
  "function_call": {},
  "stream": true,
  "temperature": 123,
  "stop": {},
  "n": 123,
  "logprobs": true,
  "seed": 123,
  "user": "<string>",
  "tools": [
    {
      "type": "<string>",
      "function": {
        "name": "<string>",
        "description": "<string>",
        "parameters": {},
        "strict": true
      }
    }
  ],
  "modalities": [
    "<string>"
  ],
  "audio": {},
  "metadata": {
    "key": "<string>"
  },
  "prediction": {},
  "verbosity": "<string>",
  "functions": [
    {}
  ]
}
'

{
  "system_fingerprint": "<string>",
  "service_tier": "<string>",
  "id": "<string>",
  "object": "<string>",
  "created": 123,
  "model": "<string>",
  "choices": [
    {
      "finish_reason": "<string>",
      "index": 123,
      "message": {
        "tool_calls": [
          {
            "id": "<string>",
            "index": 123,
            "type": "<string>",
            "function": {
              "name": "<string>",
              "arguments": "<string>"
            }
          }
        ],
        "tool_call_id": "<string>",
        "reasoning_content": "<string>",
        "content": {},
        "name": "<string>",
        "refusal": "<string>",
        "annotations": [
          {}
        ]
      },
      "logprobs": {
        "content": [
          {
            "top_logprobs": [
              {
                "token": "<string>",
                "logprob": 123,
                "bytes": [
                  123
                ]
              }
            ],
            "token": "<string>",
            "logprob": 123,
            "bytes": [
              123
            ]
          }
        ],
        "refusal": [
          {
            "top_logprobs": [
              {
                "token": "<string>",
                "logprob": 123,
                "bytes": [
                  123
                ]
              }
            ],
            "token": "<string>",
            "logprob": 123,
            "bytes": [
              123
            ]
          }
        ]
      }
    }
  ],
  "usage": {
    "prompt_tokens": 123,
    "completion_tokens": 123,
    "total_tokens": 123,
    "prompt_tokens_details": {
      "cached_tokens": 123,
      "audio_tokens": 123
    },
    "completion_tokens_details": {
      "reasoning_tokens": 123,
      "audio_tokens": 123,
      "accepted_prediction_tokens": 123,
      "rejected_prediction_tokens": 123
    },
    "ephemeral_5m_input_tokens": 123,
    "ephemeral_1h_input_tokens": 123,
    "web_search_requests": 123
  }
}

​功能特性

​使用场景

​认证方式

​快速示例

请求头

请求体

响应

功能特性

使用场景

认证方式

快速示例