> ## Documentation Index
> Fetch the complete documentation index at: https://docs.retab.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Create Experiment Run

> Create an experiment run.

The `experiment_id` and an optional `workflow_id` are supplied in the body.
When `workflow_id` is omitted, the experiment's workflow is used; when
supplied, it must match that workflow or the request is rejected with 404.

Trigger an experiment run with the current draft block configuration.

The canonical API route is flat: send `experiment_id` in the request body.
`workflow_id` is optional in the body and scopes the lookup when provided.

This is the call that produces metrics: it re-processes every experiment
document through the block with `n_consensus` parallel passes per document.
Use it after creating an experiment, after editing the block, or after
changing the document set.

The endpoint is async - it returns a run resource immediately. Poll the
experiment run with [Get Experiment Run](/api-reference/workflows/experiments/runs/get)
until it reaches a terminal status, then read metrics with [Get Experiment Run
Metrics](/api-reference/workflows/experiments/metrics/get).

Runs use the experiment's stored `n_consensus` and document set.

<RequestExample>
  ```python Python theme={null}
  from retab import Retab

  client = Retab()

  run = client.workflows.experiments.runs.create(
      workflow_id="wf_abc123",
      experiment_id="exp_abc",
  )
  print(run.id, run.lifecycle.status)

  # Wait for the run to complete, then read metrics.
  metrics = client.workflows.experiments.metrics.get(
      run.id,
      view="summary",
  )
  ```

  ```typescript TypeScript theme={null}
  import { Retab } from "@retab/node";

  const client = new Retab({ apiKey: process.env.RETAB_API_KEY });

  const run = await client.workflows.experiments.runs.create("exp_abc", "wf_abc123");
  console.log(run.id, run.lifecycle.status);

  // Wait for the run to complete, then read metrics.
  const metrics = await client.workflows.experiments.metrics.get({
    runId: run.id,
    view: "summary",
  });
  ```

  ```go Go theme={null}
  package main

  import (
  	"context"
  	"fmt"
  	"log"

  	retab "github.com/retab-dev/retab/clients/go"
  )

  func ptr[T any](v T) *T { return &v }

  func main() {
  	ctx := context.Background()

  	client, err := retab.NewClient("")
  	if err != nil {
  		log.Fatal(err)
  	}

  	run, err := client.Workflows.Experiments.Runs.Create(ctx, &retab.ExperimentRunsCreateParams{
  		WorkflowID:   ptr("wf_abc123"),
  		ExperimentID: "exp_abc",
  	})
  	if err != nil {
  		log.Fatal(err)
  	}
  	fmt.Println(run.ID, run.Lifecycle.Status())

  	// Wait for the run to complete, then read metrics.
  	metrics, err := client.Workflows.Experiments.Metrics.Get(ctx,
  		&retab.ExperimentRunMetricsGetParams{RunID: run.ID, View: ptr(retab.ExperimentRunMetricsViewSummary)})
  	if err != nil {
  		log.Fatal(err)
  	}
  	_ = metrics
  }
  ```

  ```rust Rust theme={null}
  use retab::enums::ExperimentRunMetricsView;
  use retab::models::CreateExperimentRunRequest;
  use retab::resources::experiment_run_metrics::GetParams as MetricsGetParams;
  use retab::resources::experiment_runs::CreateParams;
  use retab::Retab;

  #[tokio::main]
  async fn main() -> Result<(), Box<dyn std::error::Error>> {
      let client = Retab::new(std::env::var("RETAB_API_KEY")?);

      let mut body = CreateExperimentRunRequest::new("exp_abc");
      body.workflow_id = Some("wf_abc123".into());
      let run = client
          .workflows().experiments().runs()
          .create(CreateParams::new(body))
          .await?;
      println!("{} {:?}", run.id, run.lifecycle);

      // Wait for the run to complete, then read metrics.
      let mut params = MetricsGetParams::new(&run.id);
      params.view = Some(ExperimentRunMetricsView::Summary);
      let _metrics = client.workflows().experiments().metrics().get(params).await?;
      Ok(())
  }
  ```

  ```php PHP theme={null}
  <?php
  require 'vendor/autoload.php';

  use Retab\Client;

  $client = new Client(apiKey: getenv('RETAB_API_KEY'));

  $result = $client->workflows()->experiments()->runs()->create(
      experimentId: 'exp_abc123',
  );
  print_r($result);
  ```

  ```csharp C# theme={null}
  using Retab;
  using RetabClient = Retab.Retab;

  var apiKey = Environment.GetEnvironmentVariable("RETAB_API_KEY")!;
  var client = new RetabClient(apiKey);

  var result = await client.Workflows.Experiments.Runs.CreateAsync(new ExperimentRunsCreateOptions());
  Console.WriteLine(result);
  ```

  ```ruby Ruby theme={null}
  require 'retab'

  client = Retab::Client.new(api_key: ENV['RETAB_API_KEY'])

  result = client.workflows.experiments.runs.create
  puts result
  ```

  ```java Java theme={null}
  import com.retab.RetabClient;

  public final class Example {
    public static void main(String[] args) throws Exception {
      RetabClient client = new RetabClient(System.getenv("RETAB_API_KEY"));

      var result = client.workflows().experiments().runs().create("exp_abc123", "wf_abc123", null);
      System.out.println(result);
    }
  }
  ```

  ```curl cURL theme={null}
  # Default — use the experiment's stored n_consensus and document set.
  curl -X 'POST' \
    'https://api.retab.com/v1/workflows/experiments/runs' \
    -H 'Content-Type: application/json' \
    -H 'Authorization: Bearer <your-api-key>' \
    -d '{
      "experiment_id": "exp_abc",
      "workflow_id": "wf_abc123"
    }'

  ```
</RequestExample>

<ResponseExample>
  ```json 200 theme={null}
  {
    "id": "exprun_2",
    "workflow": {
      "workflow_id": "wf_abc123",
      "version_id": "draft_2026_05_18"
    },
    "trigger": { "type": "api" },
    "lifecycle": { "status": "pending" },
    "timing": {
      "created_at": "2026-05-18T10:00:00Z",
      "started_at": null,
      "completed_at": null
    },
    "experiment_id": "exp_abc",
    "block_id": "extract-invoice",
    "block_kind": "extract",
    "block_execution_fingerprint": "0ff93ddc7cefcb42",
    "documents_fingerprint": "ddd95baadce6045f",
    "total_document_count": 12,
    "completed_document_count": 0,
    "error_count": 0,
    "n_consensus": 5
  }
  ```
</ResponseExample>


## OpenAPI

````yaml POST /v1/workflows/experiments/runs
openapi: 3.1.0
info:
  title: FastAPI
  version: 0.1.0
servers:
  - url: https://api.retab.com
security: []
paths:
  /v1/workflows/experiments/runs:
    post:
      tags:
        - Workflows
        - Workflow Experiments
      summary: Create Experiment Run Flat
      description: >-
        Create an experiment run.


        The `experiment_id` and an optional `workflow_id` are supplied in the
        body.

        When `workflow_id` is omitted, the experiment's workflow is used; when

        supplied, it must match that workflow or the request is rejected with
        404.
      operationId: create_experiment_run
      requestBody:
        content:
          application/json:
            schema:
              $ref: '#/components/schemas/CreateExperimentRunRequest'
        required: true
      responses:
        '202':
          description: Successful Response
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/WorkflowExperimentRun'
        '422':
          description: Validation Error
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/HTTPValidationError'
      security:
        - HTTPBearer: []
components:
  schemas:
    CreateExperimentRunRequest:
      properties:
        experiment_id:
          type: string
          title: Experiment Id
          description: The experiment to create a run for.
        workflow_id:
          anyOf:
            - type: string
            - type: 'null'
          title: Workflow Id
          description: >-
            Optional. When omitted, the workflow is derived from the experiment
            record. When supplied, must match the experiment's workflow_id (404
            otherwise).
        plan_token:
          anyOf:
            - type: string
            - type: 'null'
          title: Plan Token
          description: >-
            Optional short-lived token returned by the run-plan preview. When
            supplied, run creation rejects if the current plan no longer matches
            the preview.
      additionalProperties: false
      type: object
      required:
        - experiment_id
      title: CreateExperimentRunRequest
      description: |-
        Request body to create an experiment run.

        `workflow_id` is optional; when omitted it is taken from the experiment,
        and when supplied it must match the experiment's workflow.
    WorkflowExperimentRun:
      properties:
        id:
          type: string
          title: Id
        workflow_id:
          type: string
          title: Workflow Id
        workflow_version_id:
          type: string
          title: Workflow Version Id
        trigger:
          $ref: '#/components/schemas/ExperimentRunTrigger'
        experiment_id:
          type: string
          title: Experiment Id
        block_id:
          type: string
          title: Block Id
        block_type:
          type: string
          enum:
            - extract
            - classifier
            - split
            - for_each
          title: Block Type
        n_consensus:
          type: integer
          enum:
            - 3
            - 5
            - 7
          title: N Consensus
        lifecycle:
          oneOf:
            - $ref: '#/components/schemas/PendingWorkflowExperimentRun'
            - $ref: '#/components/schemas/QueuedWorkflowExperimentRun'
            - $ref: '#/components/schemas/RunningWorkflowExperimentRun'
            - $ref: '#/components/schemas/CompletedWorkflowExperimentRun'
            - $ref: '#/components/schemas/ErrorWorkflowExperimentRun'
            - $ref: '#/components/schemas/CancelledWorkflowExperimentRun'
          title: Lifecycle
          discriminator:
            propertyName: status
            mapping:
              cancelled:
                $ref: '#/components/schemas/CancelledWorkflowExperimentRun'
              completed:
                $ref: '#/components/schemas/CompletedWorkflowExperimentRun'
              error:
                $ref: '#/components/schemas/ErrorWorkflowExperimentRun'
              pending:
                $ref: '#/components/schemas/PendingWorkflowExperimentRun'
              queued:
                $ref: '#/components/schemas/QueuedWorkflowExperimentRun'
              running:
                $ref: '#/components/schemas/RunningWorkflowExperimentRun'
        timing:
          $ref: '#/components/schemas/ExperimentRunTiming'
        parent_run_id:
          anyOf:
            - type: string
            - type: 'null'
          title: Parent Run Id
        block_version_id:
          anyOf:
            - type: string
            - type: 'null'
          title: Block Version Id
        metrics_validity_fingerprint:
          anyOf:
            - type: string
            - type: 'null'
          title: Metrics Validity Fingerprint
        metrics_validity_fingerprint_version:
          anyOf:
            - type: integer
            - type: 'null'
          title: Metrics Validity Fingerprint Version
        block_execution_fingerprint:
          type: string
          title: Block Execution Fingerprint
        documents_fingerprint:
          type: string
          title: Documents Fingerprint
        score:
          anyOf:
            - type: number
            - type: 'null'
          title: Score
        total_document_count:
          type: integer
          title: Total Document Count
          default: 0
        completed_document_count:
          type: integer
          title: Completed Document Count
          default: 0
        document_count:
          type: integer
          title: Document Count
          default: 0
        error_count:
          type: integer
          title: Error Count
          default: 0
      type: object
      required:
        - block_execution_fingerprint
        - block_id
        - block_type
        - documents_fingerprint
        - experiment_id
        - id
        - lifecycle
        - n_consensus
        - timing
        - trigger
        - workflow_id
        - workflow_version_id
      title: WorkflowExperimentRun
      description: A single execution of an experiment, identified by `id`.
    HTTPValidationError:
      properties:
        detail:
          items:
            $ref: '#/components/schemas/ValidationError'
          type: array
          title: Detail
          default: []
      type: object
      title: HTTPValidationError
    ExperimentRunTrigger:
      properties:
        type:
          anyOf:
            - type: string
            - type: 'null'
          title: Type
      type: object
      title: ExperimentRunTrigger
    PendingWorkflowExperimentRun:
      properties:
        status:
          type: string
          const: pending
          title: Status
          default: pending
      type: object
      title: PendingWorkflowExperimentRun
      description: The experiment run has been created but execution has not started.
    QueuedWorkflowExperimentRun:
      properties:
        status:
          type: string
          const: queued
          title: Status
          default: queued
      type: object
      title: QueuedWorkflowExperimentRun
      description: The experiment run is enqueued and waiting for a worker.
    RunningWorkflowExperimentRun:
      properties:
        status:
          type: string
          const: running
          title: Status
          default: running
      type: object
      title: RunningWorkflowExperimentRun
      description: The experiment run is executing.
    CompletedWorkflowExperimentRun:
      properties:
        status:
          type: string
          const: completed
          title: Status
          default: completed
      type: object
      title: CompletedWorkflowExperimentRun
      description: The experiment run finished successfully.
    ErrorWorkflowExperimentRun:
      properties:
        status:
          type: string
          const: error
          title: Status
          default: error
        message:
          type: string
          title: Message
          description: Human-readable error message
          default: (no message)
        details:
          anyOf:
            - $ref: '#/components/schemas/ErrorDetails'
            - type: 'null'
          description: Structured error context including stack trace
      type: object
      title: ErrorWorkflowExperimentRun
      description: |-
        The experiment run failed.

        Carries a human-readable `message` and a structured `details` envelope
        consumers can branch on instead of parsing free text.
    CancelledWorkflowExperimentRun:
      properties:
        status:
          type: string
          const: cancelled
          title: Status
          default: cancelled
        reason:
          anyOf:
            - type: string
            - type: 'null'
          title: Reason
          description: Human-readable reason, when known
      type: object
      title: CancelledWorkflowExperimentRun
      description: >-
        The experiment run was cancelled before reaching a natural terminal
        state.
    ExperimentRunTiming:
      properties:
        created_at:
          type: string
          format: date-time
          title: Created At
          description: When the experiment run record was created
        started_at:
          anyOf:
            - type: string
              format: date-time
            - type: 'null'
          title: Started At
        completed_at:
          anyOf:
            - type: string
              format: date-time
            - type: 'null'
          title: Completed At
        duration_ms:
          anyOf:
            - type: integer
            - type: 'null'
          title: Duration Ms
      type: object
      title: ExperimentRunTiming
    ValidationError:
      properties:
        loc:
          items:
            anyOf:
              - type: string
              - type: integer
          type: array
          title: Location
        msg:
          type: string
          title: Message
        type:
          type: string
          title: Error Type
        input:
          title: Input
          default: null
        ctx:
          type: object
          title: Context
          default: {}
      type: object
      required:
        - loc
        - msg
        - type
      title: ValidationError
    ErrorDetails:
      properties:
        message:
          anyOf:
            - type: string
            - type: 'null'
          title: Message
          description: >-
            Human-readable error message. Free-text; the structured fields below
            are the machine-readable counterpart.
        stack_trace:
          anyOf:
            - type: string
            - type: 'null'
          title: Stack Trace
          description: Full stack trace
        block_id:
          anyOf:
            - type: string
            - type: 'null'
          title: Block Id
          description: ID of the block that failed
        block_name:
          anyOf:
            - type: string
            - type: 'null'
          title: Block Name
          description: Name/label of the block that failed
        error_code:
          anyOf:
            - type: string
            - type: 'null'
          title: Error Code
          description: Error code if available
        context:
          anyOf:
            - additionalProperties: true
              type: object
            - type: 'null'
          title: Context
          description: Additional context about the error
      type: object
      title: ErrorDetails
      description: |-
        Detailed error information for debugging.

        Captures stack traces and context about where and why an error occurred.
  securitySchemes:
    HTTPBearer:
      type: http
      scheme: bearer

````