compatible Responses-Retrieve a response

Retrieve a completed model response by its Response ID.

China (Beijing)

The base_url for SDK calls is: https://dashscope.aliyuncs.com/compatible-mode/v1

HTTP request URL: GET https://dashscope.aliyuncs.com/compatible-mode/v1/responses/{response_id}

## Path parameters

**response_id **string* *(required) The Response ID to retrieve. The format is resp_xxx. You can get it from the response of the Create response API. Only Response IDs returned when store=true was set in the original creation request can be retrieved.

## Python

import os
from openai import OpenAI

client = OpenAI(
 # If you have not configured an environment variable, replace the following line with: api_key="sk-xxx"
 api_key=os.getenv("DASHSCOPE_API_KEY"),
 base_url="https://dashscope.aliyuncs.com/compatible-mode/v1",
)

response = client.responses.retrieve("resp_xxx")
print(response)

## Node.js

import OpenAI from "openai";

const openai = new OpenAI({
 // If you have not configured an environment variable, replace the following line with: apiKey: "sk-xxx"
 apiKey: process.env.DASHSCOPE_API_KEY,
 baseURL: "https://dashscope.aliyuncs.com/compatible-mode/v1"
});

async function main() {
 const response = await openai.responses.retrieve("resp_xxx");
 console.log(response);
}

main();

## curl

curl https://dashscope.aliyuncs.com/compatible-mode/v1/responses/resp_xxx \\
 -H "Authorization: Bearer $DASHSCOPE_API_KEY"

## Response

Returns the same Response object as the Create response API. The fields are described below:

{
 "background": false,
 "completed_at": 1778676420,
 "created_at": 1778676418,
 "frequency_penalty": 0.0,
 "id": "resp_801bc2c4-93d9-910f-b35d-5274f5a737c1",
 "metadata": {},
 "model": "qwen-plus",
 "object": "response",
 "output": \[
 {
 "content": \[
 {
 "annotations": \[\],
 "text": "Hello! Nice to meet you. How can I help you?",
 "type": "output_text"
 }
 \],
 "id": "msg_8c54756c-9b65-4a95-81d7-4276d91406db",
 "role": "assistant",
 "status": "completed",
 "type": "message"
 }
 \],
 "parallel_tool_calls": true,
 "presence_penalty": 0.0,
 "service_tier": "default",
 "status": "completed",
 "store": true,
 "temperature": 1.0,
 "tool_choice": "auto",
 "tools": \[\],
 "top_logprobs": 0,
 "top_p": 1.0,
 "usage": {
 "input_tokens": 45,
 "input_tokens_details": {
 "cached_tokens": 0
 },
 "output_tokens": 63,
 "output_tokens_details": {
 "reasoning_tokens": 0
 },
 "total_tokens": 108,
 "x_details": \[
 {
 "input_tokens": 45,
 "output_tokens": 63,
 "prompt_tokens_details": {
 "cached_tokens": 0
 },
 "total_tokens": 108,
 "x_billing_type": "response_api"
 }
 \]
 }
}

id *string* Unique identifier for this response, in the format resp_xxx.

object *string* Object type. Always response.

status *string* Response status. Possible values: completed, failed, in_progress, cancelled, queued, incomplete.

created_at *integer* Unix timestamp (in seconds) when the response was created.

completed_at *integer* Unix timestamp (in seconds) when the response finished generating. null if the response is not complete.

error *object* Error object returned when the model fails to generate a response. null on success.

model *string* Model ID used to generate the response.

output *array* Output items generated by the model. The type and order of elements depend on the model's response. Array element properties

type *string* Output item type. Possible values:

• message: Message containing the model's final reply.
• reasoning: Reasoning output. Returned when reasoning.effort is set to a value other than none or when thinking mode is enabled. Reasoning tokens are counted in output_tokens_details.reasoning_tokens and billed accordingly.
• function_call: Function call output. Returned when using a custom function tool. You must handle the function call and return a result.
• web_search_call: Web search call output. Returned when using the web_search tool.
• code_interpreter_call: Code execution output. Returned when using the code_interpreter tool.
• web_extractor_call: Web extraction output. Returned when using the web_extractor tool. Must be used together with the web_search tool.
• web_search_image_call: Text-to-image search call output. Returned when using the web_search_image tool. Contains a list of images found.
• image_search_call: Image-to-image search call output. Returned when using the image_search tool. Contains a list of similar images found.
• mcp_call: MCP call output. Returned when using the mcp tool. Contains the result from the MCP service.
• file_search_call: Knowledge base search call output. Returned when using the file_search tool. Contains retrieval queries and results from the knowledge base.

id *string* Unique identifier for the output item. Present for all output types.

role *string* Message role. Always assistant. Present only when type is message.

status *string* Output item status. Possible values: completed, in_progress. Present when type is not reasoning.

name *string* Tool or function name. Present when type is function_call, web_search_image_call, image_search_call, or mcp_call. For web_search_image_call and image_search_call, the value is fixed as "web_search_image" and "image_search", respectively. For mcp_call, the value is the specific function name called in the MCP service (for example, amap-maps-maps_geo).

arguments *string* Tool call arguments, formatted as a JSON string. Present when type is function_call, web_search_image_call, image_search_call, or mcp_call. Parse it using JSON.parse() before use. Contents vary by tool type:

• web_search_image_call: {"queries": \["search keyword 1", "search keyword 2"\]}. The queries field contains a list of search keywords automatically generated by the model based on user input.
• image_search_call: {"img_idx": 0, "bbox": \[0, 0, 1000, 1000\]}. The img_idx field is the index of the input image (starting from 0). The bbox field contains the bounding box coordinates [x1, y1, x2, y2] of the search area, with values ranging from 0 to 1000.
• function_call: An argument object generated according to the user-defined function parameter schema.
• mcp_call: An argument object for the function called in the MCP service.

call_id *string* Unique identifier for the function call. Present only when type is function_call. Use this ID to associate the request with the response when returning function call results.

content *array* Message content array. Present only when type is message. Array element properties

type *string* Content type. Always output_text.

text *string* Text content generated by the model.

annotations *array* Text annotations. Usually an empty array.

summary *array* Reasoning summaries. Present only when type is reasoning. Each element contains a type field (value summary_text) and a text field (the summary text).

action *object* Search action information. Present only when type is web_search_call. Properties

query *string* Search query keyword.

type *string* Search type. Always search.

sources *array* Search sources. Each element contains a type field and a url field.

code *string* Code generated and executed by the model. Present only when type is code_interpreter_call.

outputs *array* Code execution outputs. Present only when type is code_interpreter_call. Each element contains a type field (value logs) and a logs field (code execution logs).

container_id *string* Code interpreter container identifier. Present only when type is code_interpreter_call. Use it to associate multiple code executions within the same session.

goal *string* Description of the extraction goal, specifying what information to extract from the webpage. Present only when type is web_extractor_call.

output *string* Output result of the tool call, formatted as a string.

• When type is web_extractor_call, this field contains a summary of the extracted webpage content.
• When type is web_search_image_call or image_search_call, this field is a JSON string containing an array of image search results. Each result includes a title (image title), url (image URL), and index (ordinal number) field.
• When type is mcp_call, this field is a JSON string result returned by the MCP service.

urls *array* URLs of the webpages extracted. Present only when type is web_extractor_call.

server_label *string* MCP service label. Present only when type is mcp_call. Identifies the MCP service used for this call.

queries *array* Queries used for knowledge base retrieval. Present only when type is file_search_call. Each element is a search query generated by the model.

results *array* Knowledge base retrieval results. Present only when type is file_search_call. Array element properties

file_id *string* File ID of the matched document.

filename *string* Filename of the matched document.

score *float* Relevance score, ranging from 0 to 1. Higher values indicate greater relevance.

text *string* Snippet of the matched document content.

usage *object* Token usage information for this request. Properties

input_tokens *integer* Number of input tokens.

output_tokens *integer* Number of output tokens generated by the model.

total_tokens *integer* Total tokens used (input_tokens + output_tokens).

input_tokens_details *object* Breakdown of input tokens. Properties

cached_tokens *integer* Number of cached tokens.

output_tokens_details *object* Breakdown of output tokens. Properties

reasoning_tokens *integer* Number of reasoning tokens.

x_details *array* Billing details. Properties

input_tokens *integer* Input tokens for this billing type.

output_tokens *integer* Output tokens for this billing type.

total_tokens *integer* Total tokens for this billing type.

x_billing_type *string* Always response_api.

prompt_tokens_details *object* Returned when session caching is enabled. Contains a cached_tokens field (number of cached tokens).

x_tools *object* Tool usage statistics. When built-in tools are used, this field contains the call count for each tool. Example: {"web_search": {"count": 1}​}

tools *array* Echoes the tools parameter from the creation request. Structure matches the tools parameter in the request body. Empty array \[\] if no tools were used.

tool_choice *string* Echoes the tool_choice parameter from the creation request. Possible values: auto, none, required.

parallel_tool_calls *boolean* Echoes the parallel_tool_calls parameter from the creation request. Indicates whether the model can call multiple tools in parallel.

temperature *float* Echoes the temperature parameter from the creation request. Controls the diversity of model output. Valid range: [0, 2). Returns the model default if not set.

top_p *float* Echoes the top_p parameter from the creation request. Nucleus sampling probability threshold. Valid range: (0, 1.0]. Returns the model default if not set.

frequency_penalty *float* Echoes the frequency_penalty parameter from the creation request. Positive values reduce the likelihood of repeated words.

presence_penalty *float* Echoes the presence_penalty parameter from the creation request. Positive values increase the likelihood of introducing new topics.

top_logprobs *integer* Echoes the top_logprobs parameter from the creation request. Number of most likely tokens returned at each position. 0 if not enabled.

store *boolean* Echoes the store parameter from the creation request. true means the response is stored and can be referenced by previous_response_id. false means it is not stored.

service_tier *string* Service tier. Always default.

background *boolean* Whether the response was executed asynchronously. Model Studio currently supports synchronous calls only, so this is always false.

metadata *object* Echoes the metadata parameter from the creation request. Custom key-value pairs attached to the response. Empty object {} if not set.

Error response

Returned when the specified Response ID does not exist:

HELPCODEESCAPE-json
{
    "error": {
        "message": "Response with id 'resp_xxx' not found.",
        "type": "InvalidParameter"
    }
}