添加多轮上下文调用ChatGPT API时出现400 Bad Request错误

核心错误原因及修复 #

1. 缺少必填的model参数 #

OpenAI Chat Completions API要求请求必须指定使用的模型（如gpt-3.5-turbo），你的代码中Request对象未设置该字段，直接触发400错误。

修复：
创建Request实例时添加模型指定：

Request request = new Request()
{
    Model = "gpt-3.5-turbo" // 可根据需求替换为gpt-3.5-turbo-16k等模型
};

2. 消息角色（Role）序列化格式错误 #

API要求消息的role字段必须是小写字符串（"system"/"user"/"assistant"），如果你的RequestMessage类中Role属性序列化后首字母大写（如"Role":"User"），会导致API识别失败。

修复：
在模型类中添加Json序列化特性，强制输出小写字段名：

public class RequestMessage
{
    [JsonPropertyName("role")]
    public string Role { get; set; }

    [JsonPropertyName("content")]
    public string Content { get; set; }
}

public class Request
{
    [JsonPropertyName("model")]
    public string Model { get; set; }

    [JsonPropertyName("messages")]
    public RequestMessage[] Messages { get; set; }
}

额外优化点 #

1. 避免冗余上下文 #

你当前每次调用都会重复添加固定的"How are you?"测试对话，会导致上下文冗余、浪费token，建议将系统消息和初始对话只初始化一次：

// 在类构造函数中初始化系统消息（仅执行一次）
public YourChatService(string apiKey)
{
    _apiKey = apiKey;
    _items.Add(new RequestMessage()
    {
        Role = "system",
        Content = $"You are ChatGPT, a large language model trained by OpenAI. Answer as concisely as possible.\nKnowledge cutoff: 2021-09-01\nCurrent date: {DateTime.Now.ToString("yyyy-MM-dd")}"
    });
}

2. 复用HttpClient #

每次创建新HttpClient会造成资源浪费，建议改为类级别静态变量复用：

private static readonly HttpClient _httpClient = new HttpClient();

// 在构造函数中初始化请求头
public YourChatService(string apiKey)
{
    _apiKey = apiKey;
    _httpClient.DefaultRequestHeaders.Authorization = new AuthenticationHeaderValue("Bearer", _apiKey);
    _httpClient.DefaultRequestHeaders.Add("Accept", "application/json");
}

修复后的完整代码片段 #

using System.Text.Json.Serialization;

private List<RequestMessage> _items = new();
private static readonly HttpClient _httpClient = new HttpClient();
private readonly string _apiKey;
private readonly string _apiUrl = "https://api.openai.com/v1/chat/completions";

public YourChatService(string apiKey)
{
    _apiKey = apiKey;
    _httpClient.DefaultRequestHeaders.Authorization = new AuthenticationHeaderValue("Bearer", _apiKey);
    _httpClient.DefaultRequestHeaders.Add("Accept", "application/json");
    
    // 初始化系统消息
    _items.Add(new RequestMessage()
    {
        Role = "system",
        Content = $"You are ChatGPT, a large language model trained by OpenAI. Answer as concisely as possible.\nKnowledge cutoff: 2021-09-01\nCurrent date: {DateTime.Now.ToString("yyyy-MM-dd")}"
    });
}

async public Task<string> SendMessage(string message)
{
    Request request = new Request()
    {
        Model = "gpt-3.5-turbo"
    };

    List<RequestMessage> requestMessages = new List<RequestMessage>(_items);
    requestMessages.Add(new RequestMessage()
    {
        Role = "user",
        Content = message
    });

    request.Messages = requestMessages.ToArray();

    string requestData = JsonSerializer.Serialize(request);
    StringContent content = new StringContent(requestData, Encoding.UTF8, "application/json");

    HttpResponseMessage httpResponseMessage = await _httpClient.PostAsync(_apiUrl, content);

    if (httpResponseMessage.IsSuccessStatusCode)
    {
        string responseString = await httpResponseMessage.Content.ReadAsStringAsync();
        Response response = JsonSerializer.Deserialize<Response>(responseString);
        string responseText = response.Choices[0].Message.Content;

        _items.Add(new RequestMessage() { Role = "user", Content = message });
        _items.Add(new RequestMessage() { Role = "assistant", Content = responseText });

        return responseText;
    }
    else
    {
        // 读取API返回的详细错误信息，便于排查
        string errorDetails = await httpResponseMessage.Content.ReadAsStringAsync();
        return $"Error: {httpResponseMessage.StatusCode} - {httpResponseMessage.ReasonPhrase}\nDetails: {errorDetails}";
    }
}

// 修正后的模型类
public class Request
{
    [JsonPropertyName("model")]
    public string Model { get; set; }

    [JsonPropertyName("messages")]
    public RequestMessage[] Messages { get; set; }
}

public class RequestMessage
{
    [JsonPropertyName("role")]
    public string Role { get; set; }

    [JsonPropertyName("content")]
    public string Content { get; set; }
}

public class Response
{
    [JsonPropertyName("choices")]
    public Choice[] Choices { get; set; }
}

public class Choice
{
    [JsonPropertyName("message")]
    public RequestMessage Message { get; set; }
}

排查建议 #

如果仍出现400错误，务必读取API返回的errorDetails内容，OpenAI会明确标注错误原因（如角色格式错误、上下文长度超限等），帮助快速定位问题。

内容的提问来源于stack exchange，提问作者Jay Croghan