September 15, 2021

Release notes



  • You can now use the Polylines annotation when labeling videos in the Editor.

  • Now you can connect your Google Cloud (GCP) data to Labelbox via the IAM Delegated Access workflow. See these docs for instructions.

  • Visit our new Usage limits page to view your API limits based on your Labelbox tier.

  • The ability to add custom metrics to your Model Diagnostics workflow is now live. See Custom metrics basics and the Custom metrics demo to learn how to use set this up.

  • Predictions can now be filtered by custom metrics in the Models tab.

  • In the Models tab, Model Runs can now be sorted by custom scalar metrics uploaded alongside predictions.


  • You can now set a Delegated Access integration as the default setting for your Organization. See our IAM Delegated Access docs for more information.

  • You can now use the Batch selection feature to select individual Data Rows across Datasets and attach them to a project for labeling. Reach out to us at [email protected] to request access.

  • When you select a project and click on the Overview tab, you can now click on the metrics in the Labeling progress and Review sections. Doing so will lead you to filtered views for each.

  • Minor speed improvements added to Model Assisted Labeling.

Python SDK

See the full full changelog in Github for a running list of updates to our Python SDK.


  • New IAMIntegration entity

  • Client.create_dataset() compatibility with IAM Delegated Access

  • Organization.get_iam_integrations() to list all IAM Delegated Access integrations available to an organization

  • Organization.get_default_iam_integration() to only get the default IAM Delegated Access integration

  • Custom Metrics can now be included and uploaded along with Model Run predictions in the Model Diagnostics workflow.