Implementation:Openai Openai node Eval Run OutputItems

Knowledge Sources	Openai_Openai_node
Domains	SDK, Evals
Last Updated	2026-02-15 12:00 GMT

Overview

The OutputItems class is the Eval Run Output Items resource in the openai-node SDK, providing methods to retrieve and list individual output items from an evaluation run.

Description

The OutputItems class extends APIResource and wraps the /evals/{eval_id}/runs/{run_id}/output_items REST endpoints. It is accessed via client.evals.runs.outputItems and provides two methods: retrieve for fetching a single output item by ID, and list for paginating output items within a run.

Each output item (represented by OutputItemRetrieveResponse or OutputItemListResponse) contains a complete record of one data point that was evaluated during a run. The output item includes: id, created_at, datasource_item (the original input data), datasource_item_id, eval_id, run_id, object ('eval.run.output_item'), status, a results array of grader results, and a sample object.

The results array contains grader results, each with name, passed (boolean), score (numeric), optional sample data, and optional type. The sample object contains the full input/output context: input messages, output messages, model, error (an EvalAPIError), finish_reason, max_completion_tokens, seed, temperature, top_p, and usage token details.

Usage

Use this resource to inspect the detailed results of an evaluation run, including the original input data, model outputs, and grader scores for each individual data point.

Code Reference

Source Location

• Repository: openai-node
• File: src/resources/evals/runs/output-items.ts

Signature

export class OutputItems extends APIResource {
  retrieve(
    outputItemID: string,
    params: OutputItemRetrieveParams,
    options?: RequestOptions,
  ): APIPromise<OutputItemRetrieveResponse>;

  list(
    runID: string,
    params: OutputItemListParams,
    options?: RequestOptions,
  ): PagePromise<OutputItemListResponsesPage, OutputItemListResponse>;
}

Import

import OpenAI from 'openai';
// Access via client.evals.runs.outputItems

I/O Contract

Inputs

Name	Type	Required	Description
outputItemID (retrieve)	string	Yes	The ID of the output item to retrieve
eval_id (retrieve/list)	string	Yes	The ID of the evaluation
run_id (retrieve)	string	Yes	The ID of the run
runID (list)	string	Yes	The ID of the run to list output items for
order (list)	'desc'	No	Sort order by timestamp (defaults to 'asc')
status (list)	'pass'	No	Filter output items by pass/fail status
after (list)	string	No	Cursor for pagination
limit (list)	number	No	Number of items per page

Outputs

Name	Type	Description
OutputItemRetrieveResponse	OutputItemRetrieveResponse	Single output item with id, created_at, datasource_item, eval_id, run_id, results, sample, status
OutputItemListResponse	OutputItemListResponse	Paginated output item with same fields as retrieve response
results[].name	string	The name of the grader that produced this result
results[].passed	boolean	Whether the grader considered the output a pass
results[].score	number	The numeric score produced by the grader
sample.input	Array<{ content: string; role: string }>	Input messages sent to the model
sample.output	Array<{ content?: string; role?: string }>	Output messages generated by the model
sample.usage	{ cached_tokens, completion_tokens, prompt_tokens, total_tokens }	Token usage details

Usage Examples

import OpenAI from 'openai';

const client = new OpenAI();

const evalId = 'eval_abc123';
const runId = 'run_xyz789';

// List all output items for a run
for await (const item of client.evals.runs.outputItems.list(runId, {
  eval_id: evalId,
  order: 'asc',
})) {
  console.log(item.id, item.status);
  for (const result of item.results) {
    console.log(`  Grader: ${result.name}, Passed: ${result.passed}, Score: ${result.score}`);
  }
}

// List only failed output items
for await (const item of client.evals.runs.outputItems.list(runId, {
  eval_id: evalId,
  status: 'fail',
})) {
  console.log('Failed item:', item.datasource_item);
  console.log('Model output:', item.sample.output);
}

// Retrieve a specific output item
const outputItem = await client.evals.runs.outputItems.retrieve('output_item_456', {
  eval_id: evalId,
  run_id: runId,
});
console.log(outputItem.sample.input);
console.log(outputItem.sample.output);
console.log(outputItem.results);

Related Pages

• Environment:Openai_Openai_node_Node_20_Runtime