聊天对话

InfiFlow 提供标准聊天对话接口，兼容 OpenAI API 格式，支持流式（SSE）和非流式响应。通过该接口可统一调用 DeepSeek、GLM 等主流 AI 模型。

1. 请求地址

POSThttps://llm.infiflow.cn/v1/chat/completions

注意

大部分兼容 OpenAI 接口标准的 AI 工具在使用服务接入地址时，可能需要在地址后添加 /v1 或 /v1/chat/completions，具体以工具要求为准，具体请参考工具文档或联系客服获取支持。

2. 请求参数

参数名	类型	必填	默认值	描述
model	string	是	无	要使用的模型 ID，如 `deepseek-v3.2`、`glm-4` 等，可在模型广场查看完整列表
messages	array[object]	是	无	对话消息列表，每条消息包含 `role` 和 `content`（详见下方 messages 参数说明）
temperature	number	否	1.0	使用什么采样温度，介于 0 和 2 之间。较高的值（如 0.8）将使输出更加随机，而较低的值（如 0.2）将使输出更加集中和确定。建议改变此参数或 `top_p`，但不要两者都改
top_p	number	否	1.0	核采样参数，模型考虑具有 top_p 概率质量的标记的结果。0.1 意味着只考虑构成前 10% 概率质量的标记。建议改变此参数或 `temperature`，但不要两者都改
n	integer	否	1	为每个输入消息生成多少个聊天完成选项
stream	boolean	否	false	如果设置，将发送部分消息增量（SSE）。当令牌可用时，令牌将作为纯数据服务器发送事件 `data:` 发送，流由消息 `data: [DONE]` 终止
stop	string/array	否	无	API 将停止生成更多令牌的最多 4 个序列
max_tokens	integer	否	无	聊天完成时生成的最大令牌数。输入标记和生成标记的总长度受模型上下文长度的限制
presence_penalty	number	否	0	-2.0 到 2.0 之间的数字。正值会根据到目前为止是否出现在文本中来惩罚新标记，从而增加模型谈论新主题的可能性
frequency_penalty	number	否	0	-2.0 到 2.0 之间的数字。正值会根据新标记在文本中的现有频率对其进行惩罚，从而降低模型逐字重复同一行的可能性
logit_bias	object	否	无	修改指定标记出现在完成中的可能性。接受一个 JSON 对象，该对象将标记 ID 映射到从 -100 到 100 的关联偏差值。-1 到 1 之间的值会减少或增加选择的可能性；-100 或 100 会导致相关令牌的禁止或独占选择
user	string	否	无	代表您的最终用户的唯一标识符，用于监控和检测滥用行为

2.1. messages 参数说明

字段	类型	必填	描述
role	string	是	消息角色，可选值：`system`（系统设定）、`user`（用户）、`assistant`（助手）
content	string	是	消息内容

3. 请求示例

3.1. OpenAI 兼容格式

bash

curl https://llm.infiflow.cn/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer sk-xxxxxx" \
  -d '{
    "model": "deepseek-v3.2",
    "messages":[
      {"role": "system", "content": "You are a helpful assistant."},
      {"role": "user", "content": "Hello"}
    ],
    "temperature": 0.7,
    "stream": true
  }'

powershell

$body = @{
    model = "deepseek-v3.2"
    messages = @(
        @{role = "system"; content = "You are a helpful assistant."}
        @{role = "user"; content = "Hello"}
    )
    temperature = 0.7
    stream = $true
} | ConvertTo-Json

$headers = @{
    "Content-Type" = "application/json"
    "Authorization" = "Bearer sk-xxxxxx"
}

$response = Invoke-RestMethod -Uri "https://llm.infiflow.cn/v1/chat/completions" `
    -Method Post -Headers $headers -Body $body -ContentType "application/json"

$response.choices[0].message.content

python

import requests
import json

url = "https://llm.infiflow.cn/v1/chat/completions"
headers = {
    "Content-Type": "application/json",
    "Authorization": "Bearer sk-xxxxxx"
}

data = {
    "model": "deepseek-v3.2",
    "messages": [
        {"role": "system", "content": "You are a helpful assistant."},
        {"role": "user", "content": "Hello"}
    ],
    "temperature": 0.7,
    "stream": True
}

response = requests.post(url, headers=headers, json=data)
result = response.json()
print(result["choices"][0]["message"]["content"])

java

import java.net.URI;
import java.net.http.HttpClient;
import java.net.http.HttpRequest;
import java.net.http.HttpResponse;
import com.google.gson.Gson;
import com.google.gson.JsonArray;
import com.google.gson.JsonObject;

public class ChatCompletion {
    public static void main(String[] args) throws Exception {
        HttpClient client = HttpClient.newHttpClient();

        JsonObject message1 = new JsonObject();
        message1.addProperty("role", "system");
        message1.addProperty("content", "You are a helpful assistant.");

        JsonObject message2 = new JsonObject();
        message2.addProperty("role", "user");
        message2.addProperty("content", "Hello");

        JsonArray messages = new JsonArray();
        messages.add(message1);
        messages.add(message2);

        JsonObject requestBody = new JsonObject();
        requestBody.addProperty("model", "deepseek-v3.2");
        requestBody.add("messages", messages);
        requestBody.addProperty("temperature", 0.7);
        requestBody.addProperty("stream", true);

        HttpRequest request = HttpRequest.newBuilder()
            .uri(URI.create("https://llm.infiflow.cn/v1/chat/completions"))
            .header("Content-Type", "application/json")
            .header("Authorization", "Bearer sk-xxxxxx")
            .POST(HttpRequest.BodyPublishers.ofString(requestBody.toString()))
            .build();

        HttpResponse<String> response = client.send(request, HttpResponse.BodyHandlers.ofString());
        System.out.println(response.body());
    }
}

vue

<template>
  <div>
    <p>{{ response }}</p>
  </div>
</template>

<script>
import axios from "axios";

export default {
  data() {
    return {
      response: ""
    };
  },
  methods: {
    async sendMessage() {
      try {
        const res = await axios.post(
          "https://llm.infiflow.cn/v1/chat/completions",
          {
            model: "deepseek-v3.2",
            messages: [
              { role: "system", content: "You are a helpful assistant." },
              { role: "user", content: "Hello" }
            ],
            temperature: 0.7,
            stream: false
          },
          {
            headers: {
              "Content-Type": "application/json",
              "Authorization": "Bearer sk-xxxxxx"
            }
          }
        );
        this.response = res.data.choices[0].message.content;
      } catch (error) {
        console.error("请求失败", error);
      }
    }
  }
};
</script>

javascript

const url = "https://llm.infiflow.cn/v1/chat/completions";

const headers = {
  "Content-Type": "application/json",
  "Authorization": "Bearer sk-xxxxxx"
};

const body = {
  model: "deepseek-v3.2",
  messages: [
    { role: "system", content: "You are a helpful assistant." },
    { role: "user", content: "Hello" }
  ],
  temperature: 0.7,
  stream: false
};

fetch(url, {
  method: "POST",
  headers: headers,
  body: JSON.stringify(body)
})
  .then(response => response.json())
  .then(data => {
    console.log(data.choices[0].message.content);
  })
  .catch(error => {
    console.error("请求失败", error);
  });

csharp

using System;
using System.Net.Http;
using System.Text;
using System.Threading.Tasks;
using Newtonsoft.Json;
using Newtonsoft.Json.Linq;

class ChatCompletion
{
    static async Task Main(string[] args)
    {
        using (HttpClient client = new HttpClient())
        {
            client.DefaultRequestHeaders.Add("Authorization", "Bearer sk-xxxxxx");

            JObject requestBody = new JObject
            {
                { "model", "deepseek-v3.2" },
                { "messages", new JArray
                    {
                        new JObject { { "role", "system" }, { "content", "You are a helpful assistant." } },
                        new JObject { { "role", "user" }, { "content", "Hello" } }
                    }
                },
                { "temperature", 0.7 },
                { "stream", true }
            };

            string json = JsonConvert.SerializeObject(requestBody);
            HttpContent content = new StringContent(json, Encoding.UTF8, "application/json");

            HttpResponseMessage response = await client.PostAsync(
                "https://llm.infiflow.cn/v1/chat/completions", content);

            string result = await response.Content.ReadAsStringAsync();
            Console.WriteLine(result);
        }
    }
}

Imports System.Net.Http
Imports System.Text
Imports Newtonsoft.Json.Linq

Module ChatCompletion
    Sub Main()
        Dim client As New HttpClient()
        client.DefaultRequestHeaders.Add("Authorization", "Bearer sk-xxxxxx")

        Dim requestBody As New JObject()
        requestBody("model") = "deepseek-v3.2"

        Dim messages As New JArray()
        Dim msg1 As New JObject()
        msg1("role") = "system"
        msg1("content") = "You are a helpful assistant."
        messages.Add(msg1)

        Dim msg2 As New JObject()
        msg2("role") = "user"
        msg2("content") = "Hello"
        messages.Add(msg2)

        requestBody("messages") = messages
        requestBody("temperature") = 0.7
        requestBody("stream") = True

        Dim json As String = requestBody.ToString()
        Dim content As New StringContent(json, Encoding.UTF8, "application/json")

        Dim response As HttpResponseMessage = client.PostAsync(
            "https://llm.infiflow.cn/v1/chat/completions", content).Result

        Dim result As String = response.Content.ReadAsStringAsync().Result
        Console.WriteLine(result)
    End Sub
End Module

java

import System.*;
import System.Net.*;
import System.IO.*;
import System.Text.*;

public class ChatCompletion
{
    public static void main(String[] args)
    {
        try
        {
            String url = "https://llm.infiflow.cn/v1/chat/completions";
            HttpWebRequest request = (HttpWebRequest)WebRequest.Create(url);
            request.set_Method("POST");
            request.set_ContentType("application/json");
            request.get_Headers().Add("Authorization", "Bearer sk-xxxxxx");

            String jsonBody = "{\"model\":\"deepseek-v3.2\",\"messages\":[{\"role\":\"system\",\"content\":\"You are a helpful assistant.\"},{\"role\":\"user\",\"content\":\"Hello\"}],\"temperature\":0.7,\"stream\":true}";

            byte[] byteArray = Encoding.get_UTF8().GetBytes(jsonBody);
            request.set_ContentLength(byteArray.get_Length());

            Stream dataStream = request.GetRequestStream();
            dataStream.Write(byteArray, 0, byteArray.get_Length());
            dataStream.Close();

            HttpWebResponse response = (HttpWebResponse)request.GetResponse();
            StreamReader reader = new StreamReader(response.GetResponseStream());
            String result = reader.ReadToEnd();

            Console.WriteLine(result);
            reader.Close();
            response.Close();
        }
        catch (Exception ex)
        {
            Console.WriteLine("请求失败: " + ex.getMessage());
        }
    }
}

3.2. Claude 原生格式

支持 Anthropic 官方的消息流格式。

POST/v1/messages

bash

curl https://llm.infiflow.cn/v1/messages \
  -H "anthropic-version: 2023-06-01" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer sk-xxxxxx" \
  -d '{
    "model": "claude-3-opus-20240229",
    "max_tokens": 1024,
    "messages": [
      {"role": "user", "content": "Hello"}
    ]
  }'

3.3. Gemini 原生格式

兼容 Google AI Studio 的 API 结构。

POST/v1beta/models/{model}:generateContent

bash

curl https://llm.infiflow.cn/v1beta/models/{model}:generateContent \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer sk-xxxxxx" \
  -d '{
    "contents": [{
      "parts": [{"text": "Hello"}]
    }]
  }'

4. 响应示例

4.1. 非流式响应

json

{
  "id": "chatcmpl-9a8b7c6d5e4f3g2h1i0j",
  "object": "chat.completion",
  "created": 1698864000,
  "model": "deepseek-v3.2",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "Hello! How can I help you today?"
      },
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 18,
    "completion_tokens": 10,
    "total_tokens": 28
  }
}

4.2. 流式响应（SSE）

text

data: {"id":"chatcmpl-9a8b7c6d","object":"chat.completion.chunk","choices":[{"delta":{"role":"assistant"},"index":0}]}

data: {"id":"chatcmpl-9a8b7c6d","object":"chat.completion.chunk","choices":[{"delta":{"content":"Hello"},"index":0}]}

data: {"id":"chatcmpl-9a8b7c6d","object":"chat.completion.chunk","choices":[{"delta":{"content":"! How can I help you today?"},"index":0}]}

data: {"id":"chatcmpl-9a8b7c6d","object":"chat.completion.chunk","choices":[{"delta":{},"finish_reason":"stop","index":0}]}

data: [DONE]

5. 响应参数说明

参数名	类型	描述
id	string	对话的唯一标识符
object	string	对象类型，非流式时为 `chat.completion`；流式时为 `chat.completion.chunk`
created	integer	创建时间戳（Unix 时间戳）
model	string	使用的模型 ID
choices	array	对话完成选项列表
choices[].index	integer	选项索引
choices[].message	object	非流式响应下的消息对象，包含 role 和 content
choices[].delta	object	流式响应下的增量消息对象，包含 role 或 content
choices[].finish_reason	string	完成原因，可选值：`stop`（正常结束）、`length`（达到长度限制）、`null`（流式传输中）
usage.prompt_tokens	integer	提示词消耗的 token 数量
usage.completion_tokens	integer	补全消耗的 token 数量
usage.total_tokens	integer	总消耗 token 数量

6. 常见错误码

状态码	描述	排查建议
400	请求参数错误	检查 model 和 messages 参数是否正确
401	认证失败	检查 API 密钥是否有效，可在控制中心的API密钥页面重新生成
429	请求频率过高	降低请求频率，或前往账户充值提升调用配额
500	服务端内部错误	请稍后重试，如持续出现请联系技术支持

注意事项：

流式响应（stream=true）使用 Server-Sent Events（SSE）协议，客户端需逐行解析 data: 前缀的数据
建议设置合理的 temperature 值，对话任务推荐 0.7 左右，代码生成任务推荐 0.2 左右
所有请求均需在 Header 中携带有效的 Bearer Token，请妥善保管您的 API 密钥
可在费用明细页面查看 API 调用的详细消耗记录

聊天对话 ​

1. 请求地址 ​

2. 请求参数 ​

2.1. messages 参数说明 ​

3. 请求示例 ​

3.1. OpenAI 兼容格式 ​

3.2. Claude 原生格式 ​

3.3. Gemini 原生格式 ​

4. 响应示例 ​

4.1. 非流式响应 ​

4.2. 流式响应（SSE） ​

5. 响应参数说明 ​

6. 常见错误码 ​

聊天对话

1. 请求地址

2. 请求参数

2.1. messages 参数说明

3. 请求示例

3.1. OpenAI 兼容格式

3.2. Claude 原生格式

3.3. Gemini 原生格式

4. 响应示例

4.1. 非流式响应

4.2. 流式响应（SSE）

5. 响应参数说明

6. 常见错误码