在Android Kotlin项目中集成ChatGPT API遇404错误求助

我想要在Android的简单Activity中集成ChatGPT API，目前遇到无法获取响应的问题，Logcat中显示错误信息“Unexpected response code: 404”。我的Activity包含用于展示响应的TextView、输入问题的EditText以及触发接口调用的Button，以下是我编写的getResponse函数：

private fun getResponse(question: String, callback: (String) -> Unit) {
        val apiKey = "my api key"
        val url = "https://api.openai.com/v1/completions"

        val requestBody = """
            {
            "model": "text-davinci-003",
            "prompt": "$question",
            "max_tokens": 500,
            "temperature": 0
            }
        """.trimIndent()

        val request = Request.Builder()
            .url(url)
            .addHeader("Content-Type", "application/json")
            .addHeader("Authorization", "Bearer $apiKey")
            .post(requestBody.toRequestBody("application/json".toMediaTypeOrNull()))
            .build()

        client.newCall(request).enqueue(object : Callback {
            override fun onFailure(call: Call, e: IOException) {
                Log.e("error", "API request failed", e)
                callback("Failed to get a response")
            }

            override fun onResponse(call: Call, response: Response) {
                response.use {
                    if (!it.isSuccessful) {
                        Log.e("error", "Unexpected response code: ${it.code}")
                        callback("Failed to get a response")
                        return
                    }

                    val body = it.body?.string()
                    if (body == null) {
                        Log.e("error", "Response body is null")
                        callback("Failed to get a response")
                        return
                    }

                    Log.v("data", body)

                    try {
                        val jsonObject = JSONObject(body)
                        val jsonArray: JSONArray = jsonObject.getJSONArray("choices")
                        val textResult = jsonArray.getJSONObject(0).getString("text")
                        callback(textResult.trim())
                    } catch (e: Exception) {
                        Log.e("error", "Failed to parse JSON", e)
                        callback("Failed to parse response")
                    }
                }
            }
        })
    }

排查与解决方案 #

• 确认模型与端点匹配：text-davinci-003属于已归档的旧模型，部分新API密钥可能无法访问该模型。建议替换为当前可用的gpt-3.5-turbo-instruct模型，端点仍为https://api.openai.com/v1/completions，确保模型名称拼写正确。
• 验证API密钥有效性：检查密钥是否在OpenAI后台处于活跃状态，是否有足够调用额度，且未被限制访问Completion接口。
• 检查应用网络权限：在AndroidManifest.xml中确认添加了网络权限：

  <uses-permission android:name="android.permission.INTERNET" />
  

• 修复请求体构建方式：手动拼接JSON容易因特殊字符（如引号、换行）导致格式错误，建议用序列化库构建请求体，示例如下：

  // 定义数据类
  data class CompletionRequest(
      val model: String,
      val prompt: String,
      val max_tokens: Int,
      val temperature: Double
  )
  
  // 替换手动拼接的requestBody部分
  val requestBody = Gson().toJson(CompletionRequest(
      model = "gpt-3.5-turbo-instruct",
      prompt = question,
      max_tokens = 500,
      temperature = 0.0
  )).toRequestBody("application/json".toMediaTypeOrNull())
  

• 测试网络连通性：用Postman或浏览器测试https://api.openai.com/v1/completions是否能正常访问，排除地区网络限制或代理问题。

内容的提问来源于stack exchange，提问作者user24936837