Update existing examples in a dataset

Updates existing dataset examples by matching their ‘id’ field. If the ‘id’ field is not passed in the examples data, they will be rejected.

If an example ID does not match any existing example in the dataset version, it will be ignored. In other words, only examples with IDs that already exist will be updated. To add new examples, use the Insert Dataset Examples endpoint.

Adding columns that do not exist in the dataset schema is allowed, but removing existing columns is not.

Optionally, the update can create a new version of the dataset. In this case, the outcome of the update will be reflected only in the new version, while the previous version remains unchanged. If a new version is not created, the updates will be applied directly (in place) to the specified version.

Example payload

• Each item in examples[] may contain any user-defined fields.
• Each item in examples[] must include the id field to identify the example to update.
• Do not include system-managed fields on input: created_at, updated_at. Requests that contain these fields in any example will be rejected.
• Each example must contain at least one property (i.e., {} is invalid).

Valid example (create)

{
  "examples": [
    { "id": "ex_001", "question": "What is 2+2?", "answer": "4", "topic": "math" }
  ]
}

Invalid example (contains system fields)

{
  "examples": [
    { "input": "Hello" }   // ❌ 'id' not specified for update
  ]
}

⚠️ Beta Warning: This endpoint is in beta, read more here.