# 即時 Perplexity LLM 回應
**API:** Data API
**Endpoint:** `POST /v1/beta/dataforseo/ai_optimization/perplexity/llm_responses/live`
**Tags:** DataForSEO
向 Perplexity 模型傳送 `user_prompt` 並同步回傳其回覆。`model_name` 必須來自 `get_dataforseo_ai_perplexity_llm_responses_models`——`sonar`、`sonar-pro` 與 `sonar-reasoning-pro`。`web_search_country_iso_code` 可限定搜尋範圍，其他三家廠商不提供此功能。回傳 `model_name`、`input_tokens`、`output_tokens`、`reasoning_tokens`、`web_search`、`money_spent`、`datetime`、`items` 與 `fan_out_queries`。讀取回應中的 `money_spent` 即可得知一次提示詞的實際成本。採用 DataForSEO 的信封結構：資料位於 `tasks[0].result`，結果狀態位於 `tasks[0].status_code`——被拒絕的請求仍會回傳 HTTP 200。
### 計費參考
- 計量方式：依回報的回應成本計量。
- 參考請求估算：0.0163 美元。
- 參考客戶倍率：1.0。不得對已結算的客戶金額再次套用。
僅供參考的中繼資料，並非帳戶報價、保證價格或消費上限。資源額度並非 Felo 額度。最終費用依 Felo 計費記錄為準。
## Base URL
- `https://openapi.felo.ai`
## Authentication
- `FeloBearerAuth` (http bearer, bearer format Felo API Key)
- `FeloApiKey` (apiKey in header `X-API-Key`)
## Request body
Required: **no**
### application/json
Schema:
```json
{
"type": "array",
"items": {
"type": [
"object",
"null"
],
"oneOf": [
{
"type": "object",
"properties": {
"user_prompt": {
"type": "string",
"description": "AI 模型的提示詞
必填欄位
要傳送給 AI 模型的問題或任務；
在 user_prompt 欄位中最多可指定 500 個字元"
},
"model_name": {
"type": "string",
"description": "AI 模型名稱
必填欄位
model_name 由實際模型名稱與版本名稱組成；
如果只指定基礎模型名稱，將預設設定其最新版本；
可個別向以下端點發出請求，取得可用的 LLM 模型清單：https://api.dataforseo.com/v3/ai_optimization/perplexity/llm_responses/models"
},
"max_output_tokens": {
"type": [
"integer",
"null"
],
"description": "AI 回應中的最大 token 數
選填欄位
最小值：1
最大值：4096；
預設值：2048；
注意：如果請求中指定的是推理模型，輸出 token 數可能超過指定的 max_output_tokens 上限"
},
"temperature": {
"type": [
"number",
"null"
],
"description": "AI 回應的隨機性
選填欄位
值越高，輸出越多元
值越低，輸出越集中
最小值：0
最大值：1.9
預設值：0.77"
},
"top_p": {
"type": [
"number",
"null"
],
"description": "AI 回應的多樣性
選填欄位
透過限制 token 選擇來控制回應的多樣性
最小值：0
最大值：1
預設值：0.9"
},
"web_search_country_iso_code": {
"type": [
"string",
"null"
],
"description": "網路搜尋本地化用的國家代碼
選填欄位
指定國家 ISO 代碼以取得本地化的網路搜尋結果
注意：僅適用於 Perplexity Sonar 模型
範例：US"
},
"system_message": {
"type": [
"string",
"null"
],
"description": "AI 行為指令
選填欄位
定義 AI 的角色、語氣或特定行為
在 system_message 欄位中最多可指定 500 個字元"
},
"message_chain": {
"type": [
"array",
"null"
],
"items": {
"type": [
"object",
"null"
],
"oneOf": [
{
"type": "object",
"properties": {
"role": {
"type": [
"string",
"null"
],
"description": "訊息來源使用者的角色"
},
"message": {
"type": [
"string",
"null"
],
"description": "訊息文字"
}
}
}
]
},
"description": "對話歷史\n選填欄位\n代表先前對話輪次的訊息物件陣列；\n每個物件必須包含：\nrole 字串，值為 user 或 ai 角色；\nmessage 字串，包含訊息內容（最多 500 個字元）；\n陣列中最多可指定 10 個訊息物件；\n注意：對於 Perplexity 模型，訊息必須嚴格在 user 與 AI 角色之間交替（user → ai）；\n範例：\n\"message_chain\": [{\"role\":\"user\",\"message\":\"Hello, what’s up?\"},{\"role\":\"ai\",\"message\":\"Hello! I’m doing well, thank you. How can I assist you today?\"}]"
},
"tag": {
"type": [
"string",
"null"
],
"description": "使用者自訂任務識別碼
選填欄位
字元數上限為 255
可使用此參數識別任務，並將其與結果對應
可在回應的 data 物件中找到指定的 tag 值"
}
},
"example": [
{
"system_message": "communicate as if we are in a business meeting",
"message_chain": [
{
"role": "user",
"message": "Hello, what’s up?"
},
{
"role": "ai",
"message": "Hello! I’m doing well, thank you. How can I assist you today? Are there any specific topics or projects you’d like to discuss in our meeting?"
}
],
"max_output_tokens": 200,
"temperature": 0.3,
"top_p": 0.5,
"web_search_country_iso_code": "FR",
"model_name": "sonar",
"user_prompt": "provide information on how relevant the amusement park business is in France now"
}
]
}
]
},
"example": [
{
"system_message": "communicate as if we are in a business meeting",
"message_chain": [
{
"role": "user",
"message": "Hello, what’s up?"
},
{
"role": "ai",
"message": "Hello! I’m doing well, thank you. How can I assist you today? Are there any specific topics or projects you’d like to discuss in our meeting?"
}
],
"max_output_tokens": 200,
"temperature": 0.3,
"top_p": 0.5,
"web_search_country_iso_code": "FR",
"model_name": "sonar",
"user_prompt": "provide information on how relevant the amusement park business is in France now"
}
]
}
```
## Responses
### 200 — 操作成功
Headers:
- `X-Request-Id` (string) — 聯絡 Felo 支援時需提供的請求識別碼。
**application/json**
Schema: object | null
```json
{
"type": [
"object",
"null"
],
"oneOf": [
{
"$ref": "#/components/schemas/Response36Afe1097B20_AiOptimizationPerplexityLlmResponsesLiveResponseInfo"
}
]
}
```
### 400 — 無法處理該要求。請檢查請求參數。
### 401 — 需要有效的 Felo API 金鑰。
### 402 — 由於未符合計費要求，要求無法繼續。
### 403 — 該帳戶無權執行此操作。
### 429 — 已達到要求或支出上限。
### 502 — 服務無法完成該要求。
### 503 — API 或計費服務暫時無法使用。
### 504 — 服務在處理要求時逾時。
### default — 錯誤回應；上游透傳回應可能採用供應商格式
**application/json**
Schema:
```json
{}
```
## Schema definitions
### Response36Afe1097B20_AiOptimizationPerplexityLlmResponsesLiveResponseInfo
`#/components/schemas/Response36Afe1097B20_AiOptimizationPerplexityLlmResponsesLiveResponseInfo`
```json
{
"type": "object",
"allOf": [
{
"$ref": "#/components/schemas/Response36Afe1097B20_BaseResponseInfo"
},
{
"type": "object",
"properties": {
"tasks": {
"type": [
"array",
"null"
],
"items": {
"type": [
"object",
"null"
],
"oneOf": [
{
"$ref": "#/components/schemas/Response36Afe1097B20_AiOptimizationPerplexityLlmResponsesLiveTaskInfo"
}
]
},
"description": "任務陣列"
}
}
}
]
}
```
### Response36Afe1097B20_AiOptimizationPerplexityLlmResponsesLiveResultInfo
`#/components/schemas/Response36Afe1097B20_AiOptimizationPerplexityLlmResponsesLiveResultInfo`
```json
{
"type": "object",
"properties": {
"model_name": {
"type": [
"string",
"null"
],
"description": "所用 AI 模型的名稱"
},
"input_tokens": {
"type": [
"integer",
"null"
],
"description": "輸入中的 token 數
處理的 token 總數"
},
"output_tokens": {
"type": [
"integer",
"null"
],
"description": "輸出中的 token 數
AI 回應中產生的 token 總數"
},
"reasoning_tokens": {
"type": [
"integer",
"null"
]
},
"web_search": {
"type": [
"boolean",
"null"
],
"description": "表示是否使用了網路搜尋
注意：在 Perplexity Sonar 模型中，網路搜尋預設為啟用"
},
"money_spent": {
"type": [
"number",
"null"
],
"description": "AI token 成本（美元）
第三方 AI 模型供應商依其定價收取的費用"
},
"datetime": {
"type": [
"string",
"null"
],
"description": "收到結果的日期與時間
採用 UTC 格式：“yyyy-mm-dd hh-mm-ss +00:00”
範例：
2019-11-15 12:57:46 +00:00"
},
"items": {
"type": [
"array",
"null"
],
"items": {
"type": [
"object",
"null"
],
"oneOf": [
{
"$ref": "#/components/schemas/Response36Afe1097B20_MessageAiOptimizationLlmResponseElementItem"
}
]
},
"description": "回應項目陣列
包含結構化的 AI 回應資料"
},
"fan_out_queries": {
"type": [
"object",
"null"
],
"description": "扇出查詢陣列
包含從主查詢衍生的相關搜尋查詢，以提供更全面的回答"
}
}
}
```
### Response36Afe1097B20_AiOptimizationPerplexityLlmResponsesLiveTaskInfo
`#/components/schemas/Response36Afe1097B20_AiOptimizationPerplexityLlmResponsesLiveTaskInfo`
```json
{
"type": "object",
"allOf": [
{
"$ref": "#/components/schemas/Response36Afe1097B20_BaseResponseTaskInfo"
},
{
"type": "object",
"properties": {
"result": {
"type": [
"array",
"null"
],
"items": {
"type": [
"object",
"null"
],
"oneOf": [
{
"$ref": "#/components/schemas/Response36Afe1097B20_AiOptimizationPerplexityLlmResponsesLiveResultInfo"
}
]
},
"description": "結果陣列"
}
}
}
]
}
```
### Response36Afe1097B20_AnnotationInfo
`#/components/schemas/Response36Afe1097B20_AnnotationInfo`
```json
{
"type": "object",
"properties": {
"title": {
"type": [
"string",
"null"
],
"description": "引用來源的網域或標題"
},
"url": {
"type": [
"string",
"null"
],
"description": "指向引用來源的重新導向 URL
包含指向原始來源的 Vertex AI 重新導向"
},
"direct_url": {
"type": [
"string",
"null"
],
"description": "引用來源的直接 URL
包含 `url` 欄位中的 Vertex AI 重新導向所指向的原始來源 URL"
},
"start_index": {
"type": [
"integer",
"null"
],
"description": "註解索引的開始位置"
},
"end_index": {
"type": [
"integer",
"null"
],
"description": "註解索引的結束位置"
},
"text": {
"type": [
"string",
"null"
],
"description": "推理鏈章節的文字
概述模型思考過程的推理鏈章節文字"
}
}
}
```
### Response36Afe1097B20_BaseAiOptimizationLlmResponseElementItem
`#/components/schemas/Response36Afe1097B20_BaseAiOptimizationLlmResponseElementItem`
```json
{
"type": "object",
"properties": {
"type": {
"type": [
"string",
"null"
],
"description": "元素類型"
}
},
"additionalProperties": false,
"discriminator": {
"propertyName": "type",
"mapping": {
"reasoning": "#/components/schemas/Response36Afe1097B20_ReasoningAiOptimizationLlmResponseElementItem",
"message": "#/components/schemas/Response36Afe1097B20_MessageAiOptimizationLlmResponseElementItem"
}
}
}
```
### Response36Afe1097B20_BaseResponseInfo
`#/components/schemas/Response36Afe1097B20_BaseResponseInfo`
```json
{
"properties": {
"version": {
"type": [
"string",
"null"
],
"description": "API 的目前版本"
},
"status_code": {
"type": [
"integer",
"null"
],
"description": "一般狀態碼\n完整回應碼清單請見此處"
},
"status_message": {
"type": [
"string",
"null"
],
"description": "一般資訊訊息\n完整的一般資訊訊息清單請見此處"
},
"time": {
"type": [
"string",
"null"
],
"description": "總執行時間，秒"
},
"cost": {
"type": [
"number",
"null"
],
"description": "任務總費用，美元",
"format": "double"
},
"tasks_count": {
"type": [
"integer",
"null"
],
"description": "tasks 陣列中的任務數",
"format": "int64"
},
"tasks_error": {
"type": [
"integer",
"null"
],
"description": "tasks 陣列中傳回錯誤的任務數",
"format": "int64"
}
}
}
```
### Response36Afe1097B20_BaseResponseTaskInfo
`#/components/schemas/Response36Afe1097B20_BaseResponseTaskInfo`
```json
{
"properties": {
"id": {
"type": [
"string",
"null"
],
"description": "任務識別碼\n我們系統中 UUID 格式的唯一任務識別碼"
},
"status_code": {
"type": [
"integer",
"null"
],
"description": "任務的狀態碼\n由 DataForSEO 產生，範圍如下：10000-60000\n完整回應碼清單請見此處"
},
"status_message": {
"type": [
"string",
"null"
],
"description": "任務的資訊訊息\n完整的一般資訊訊息清單請見此處"
},
"time": {
"type": [
"string",
"null"
],
"description": "執行時間，秒"
},
"cost": {
"type": [
"number",
"null"
],
"description": "任務總費用，美元",
"format": "double"
},
"result_count": {
"type": [
"integer",
"null"
],
"description": "result 陣列中的元素數",
"format": "int64"
},
"path": {
"type": [
"array",
"null"
],
"items": {
"type": [
"string",
"null"
]
},
"description": "URL 路徑"
},
"data": {
"type": [
"object",
"null"
],
"additionalProperties": {
"type": [
"object",
"null"
]
},
"description": "包含你在 POST 請求中指定的相同參數"
}
}
}
```
### Response36Afe1097B20_LlmMessageSectionInfo
`#/components/schemas/Response36Afe1097B20_LlmMessageSectionInfo`
```json
{
"type": "object",
"properties": {
"type": {
"type": [
"string",
"null"
],
"description": "元素類型"
},
"text": {
"type": [
"string",
"null"
],
"description": "推理鏈章節的文字
概述模型思考過程的推理鏈章節文字"
},
"annotations": {
"type": [
"array",
"null"
],
"items": {
"type": [
"object",
"null"
],
"oneOf": [
{
"$ref": "#/components/schemas/Response36Afe1097B20_AnnotationInfo"
}
]
},
"description": "用於產生回應的參照陣列
若 web_search 參數未設為 true，則為 null
注意：即使 web_search 為 true，annotations 也可能傳回空值，因為 AI 會嘗試擷取網路資訊，但可能找不到相關結果"
}
}
}
```
### Response36Afe1097B20_MessageAiOptimizationLlmResponseElementItem
`#/components/schemas/Response36Afe1097B20_MessageAiOptimizationLlmResponseElementItem`
```json
{
"type": "object",
"allOf": [
{
"$ref": "#/components/schemas/Response36Afe1097B20_BaseAiOptimizationLlmResponseElementItem"
},
{
"type": "object",
"properties": {
"sections": {
"type": [
"array",
"null"
],
"items": {
"type": [
"object",
"null"
],
"oneOf": [
{
"$ref": "#/components/schemas/Response36Afe1097B20_LlmMessageSectionInfo"
}
]
},
"description": "推理鏈章節
包含 LLM 產生的推理鏈章節之物件陣列"
}
}
}
]
}
```