Imaginemos que cargar el modelo puede **tomar bastante tiempo**, porque tiene que leer muchos **datos del disco**. Entonces no quieres hacerlo para cada request.

Podrías cargarlo en el nivel superior del módulo/archivo, pero eso también significaría que **cargaría el modelo** incluso si solo estás ejecutando una simple prueba automatizada, entonces esa prueba sería **lenta** porque tendría que esperar a que el modelo se cargue antes de poder ejecutar una parte independiente del código.

        Doc(
            """
            Pydantic's `include` parameter, passed to Pydantic models to set the
            fields to include.
            """
        ),
    ] = None,
    exclude: Annotated[
        IncEx | None,
        Doc(
            """
            Pydantic's `exclude` parameter, passed to Pydantic models to set the
            fields to exclude.
            """
        ),
    ] = None,

@SuppressWarnings("module")
module okhttp3.dnsoverhttps {
  requires okhttp3;
  exports okhttp3.dnsoverhttps;

under the License.
-->

<project>
  <modelVersion>4.0.0</modelVersion>
  <artifactId>aid</artifactId>
  <groupId>gid</groupId>
  <version>0.1</version>
  <packaging>pom</packaging>

  <modules>
    <module> </module>
  </modules>

    /**
     * Gets the raw model to build. If not set, model source will be used to load raw model.
     *
     * @return The raw model to build or {@code null} if not set.
     */
    Model getRawModel();

    /**
     * Set raw model.
     *
     * @param rawModel
     */
    ModelBuildingRequest setRawModel(Model rawModel);

    /**
     * Gets the source of the POM to process.

				})
			}
		}
	})

	var models []User
	if err := DB.Where("name in (?)", []string{"find"}).Find(&models).Error; err != nil || len(models) != 3 {
		t.Errorf("errors happened when query find with in clause: %v, length: %v", err, len(models))
	} else {
		for idx, user := range users {
			t.Run("FindWithInClause#"+strconv.Itoa(idx+1), func(t *testing.T) {
				CheckUser(t, models[idx], user)
			})
		}
	}

You can see those schemas because they were declared with the models in the app.

That information is available in the app's **OpenAPI schema**, and then shown in the API docs.

That same information from the models that is included in OpenAPI is what can be used to **generate the client code**.

### Hey API { #hey-api }

    hashed_password = user_dict["hashed_password"],
)
```

/// info | Información

Para una explicación más completa de `**user_dict` revisa en [la documentación para **Extra Models**](../extra-models.md#about-user-in-dict).

///

## Devolver el token { #return-the-token }

El response del endpoint `token` debe ser un objeto JSON.

///

/// note

Notice that the input model is still validated.

So, if you want to receive partial updates that can omit all the attributes, you need to have a model with all the attributes marked as optional (with default values or `None`).

To distinguish from the models with all optional values for **updates** and models with required values for **creation**, you can use the ideas described in [Extra Models](extra-models.md).

### `orjson` ou Modelo de Resposta { #orjson-or-response-model }

Se o que você procura é performance, provavelmente é melhor usar um [Modelo de Resposta](../tutorial/response-model.md) do que uma resposta com `orjson`.