# get_scoring_model (/mcp/get-scoring-model)



Shows one scoring model, as [`GET /v1/scoring_models/{model_id}`](/reference/get-scoring-model). Read-only.

| Input      | Description                              |
| ---------- | ---------------------------------------- |
| `model_id` | Required. The scoring model's `model_id` |
| `api_key`  | `/mcp/keys` only                         |

Returns the fields [`list_scoring_models`](/mcp/list-scoring-models) gives, and `questions`: the ones the scoring model uses, strongest first. Each has:

* **`id`,** such as `q007`, to remove it with [`remove_question`](/mcp/remove-question) or [`calibrate`](/mcp/calibrate).
* **`text`,** the question Grayson is asked about each record.
* **`source`:** `finic` for a question Finic wrote, `user` for one the user added.
* **`weight`:** how much it moves the score, as of the latest calibration, or Finic's estimate if `calibrated_at` is `null`. Positive when a likely yes raises the score, negative when it lowers it; the larger, the more it counts.

The user's own questions it doesn't use are listed on its page in the Finic Portal. See [Questions and weights](/concepts/scoring-models#questions-and-weights).