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

# Run a kept eval set

> Start a run of a kept eval set, or score the next cases of a run in progress, up to 10 at a time.

Without `run_id` this starts a run of the set. With one it scores that run's next cases. Each call scores up to 10 cases exactly as [an unstored eval](https://docs.mithunai.com/api-reference/assistants/run-assistant-eval.md) does, and keeps them. Each case counts as one answer against your [monthly allowance](https://docs.mithunai.com/analyze/usage-and-limits.md), and the call is audited as `assistant.eval_run`. Call again with the run's `id` until `status` is no longer `running`. Requires the owner, admin or editor role.

> [!NOTE]
>
> Off by default. These routes answer `404` unless the deployment has switched on
> `ARUKZ_EVAL_HISTORY`. See [Answer quality
> evals](https://docs.mithunai.com/assistants/answer-quality.md#kept-sets-and-runs-beta).

**assistant\_id** `string` *required*

The assistant (UUID). Another organization's assistant returns `404`.

**set\_id** `string` *required*

The kept set to run. One that does not exist returns `404`.

**run\_id** `string`

The run to continue, at most 64 characters. Absent starts a new run. A run that is already
finished returns `409`.

## Response

The run so far, with every case scored.

**id** `string` *required*

The run.

**eval\_set\_id** `string` *required*

The set it ran.

**status** `string` *required*

`running`, `done` or `cannot_run`. Treat a value you do not recognise as `cannot_run`.

**problem** `string | null` *required*

Why it stopped, when it did.

**total** `integer` *required*

Cases in the set.

**scored** `integer | null` *required*

Cases scored so far.

**passed** `integer` *required*

Cases that passed.

**started\_by** `string` *required*

Who started it.

**started\_at** `string` *required*

When, RFC 3339.

**finished\_at** `string | null` *required*

When it finished, or `null` while it is running.

**results** `object[]` *required*

Each scored case, in order, as in [Run an eval](https://docs.mithunai.com/api-reference/assistants/run-assistant-eval.md).


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.