# UiPath Helix Extractor 2.0

> Configure UiPath Helix Extractor 2.0, a fine-tunable multimodal extraction model that processes document text and visual layout for improved accuracy on complex and semi-structured documents.

UiPath Helix Extractor 2.0 is a fine-tunable document extraction model for Modern projects. Built on a multimodal architecture, it processes both document text and visual layout, enabling more accurate extraction for complex and semi-structured documents.

## Availability

UiPath Helix Extractor 2.0 is generally available for tenants in **Europe** and the **United States**. You can be onboarded to this functionality if you are part of the **United States delayed (GxP) region**.

## Default behavior by region

Helix Extractor 2.0 has different default behaviors depending on your tenant's region:

* **United States, including the delayed (GxP) region**: Helix Extractor 2.0 is the default for new projects. You can revert to Helix Extractor 1.0 from the Document Type Manager until July 15, 2026 (August 21, 2026 for the delayed region), after which Helix Extractor 1.0 training is removed.
* **Europe**: Helix Extractor 2.0 remains opt-in through a **Settings** toggle, available on all projects, both new and existing.

## Improvements over Helix Extractor 1.0

| Capability | Helix Extractor 1.0 | Helix Extractor 2.0 |
|---|---|---|
| Architecture | Text-based | Multimodal — processes text and visual layout |
| Language support | Latin script only | Latin and non-Latin scripts |
| Table extraction | Standard | Improved row and column accuracy |
| Confidence alignment | Standard | Higher correlation with actual extraction accuracy |

## Enable Helix Extractor 2.0

You can enable Helix Extractor 2.0 in the following ways:
* When creating a new Modern project.
* From an existing Modern project.

**When creating a new Modern project**

1. In the **Create project** dialog, select **Modern**.
2. Enable the **Enable Helix Extractor 2.0** toggle.
3. Select **Create**.

   ![Screenshot of the Create project dialog with the Enable Helix Extractor 2.0 toggle turned on.](https://dev-assets.cms.uipath.com/assets/images/document-understanding/helix-extractor-2-0-toggle-ga-eu-623501e8.webp)

All document types added to the project use Helix Extractor 2.0 by default.

**From an existing Modern project**

1. Open **Project Settings**.
2. Under **Extraction**, enable the **Enable Helix Extractor 2.0** toggle.
3. Select **Save**.

   ![Screenshot of Project settings with the Enable Helix Extractor 2.0 toggle turned on under the Extraction section.](https://dev-assets.cms.uipath.com/assets/images/document-understanding/project-settings-helix-2-0-ga-eu-18476b86.webp)

:::note
After enabling Helix Extractor 2.0:
* Newly created document types use Helix Extractor 2.0 for training.
* All existing extractors will need to be retrained for the new setting to take effect for them.
* Document types with pinned model versions in existing project versions are not affected and continue using their original model version.
:::

## Verify the extraction model version

To confirm which model version a document type uses:

1. Create a new project version.
2. Select **Edit version**.
3. Review the model version associated with each document type.

:::note
Model versions **25.9.x** and **26.x** correspond to UiPath Helix Extractor 2.0.
:::

## Select the model type per document type

When you enable Helix Extractor 2.0 at the project level, it becomes the default for all document types. You can still configure the model type individually per document type:

1. Open **Document Type Manager**.
2. Select a document type.
3. Open **Settings**.
4. From the **Model type** drop-down list, select **Helix Extractor 2.0** or **Legacy**.
5. Select **Save**.

   ![Screenshot of the Settings tab in Document Type Manager showing the Model type drop-down list with Helix Extractor 2.0 selected.](https://dev-assets.cms.uipath.com/assets/images/document-understanding/settings-helix-extractor-2-0-cd5b27a6.webp)

:::note
Saving the change prompts the user with the option to trigger a new training event for that document type.
:::

This lets you use Helix Extractor 2.0 for document types where you want improved performance, while keeping the Legacy model for document types that already meet your automation goals.

## Known limitations

* UiPath Helix Extractor 2.0 processes a maximum of 100 pages per document at runtime. For the full list of specifications, see [Known limitations](https://docs.uipath.com/document-understanding/automation-cloud/latest/user-guide/known-limitations).
* UiPath Helix Extractor 2.0 does not support document tags (batches) yet. They will be missing from the Advanced Metrics report file.