Get Evaluation Results

Status, progress, per-config scorecards, billing and per-case results for an eval. `scorecards` is `{}` and `billing` is `null` until the eval is terminal (`completed`, `completed_with_errors`, or `failed`). Items are paged by `items_limit` (default 100, max 500) and `items_cursor`; each item carries a result object per config name. Recall, MRR and nDCG are over scored units only; a failed query is counted in `failed`, never scored as a miss. Evals of a deleted collection return `404`. Metric definitions are on the [Evaluations](/guides/evaluations) guide.

Path parameters

eval_idstringRequired

Query parameters

items_limitintegerOptional1-500Defaults to 100
items_cursorstring or nullOptional<=256 characters

Response

The eval, with items paged.
collection_namestring
configslist of objects
created_atstring
eval_idstring
idempotency_keystring
items_pageobject
previewobject
progressobject
request_idstring
statusenum
updated_atstring
upload_idstring
billingobject or nullOptional
completed_atstring or nullOptional
environmentstring or nullOptional
error_codestring or nullOptional
error_messagestring or nullOptional
itemslist of objectsOptional
scorecardsmap from strings to objectsOptional
started_atstring or nullOptional

Errors

400
Bad Request Error
401
Unauthorized Error
403
Forbidden Error
404
Not Found Error
© 2026 Captain