- Release Notes
- General Availability
- Getting Started
- Data Labeling
- ML Packages
- Out of the Box Packages
- AI Computer Vision
- Communications Mining
- UiPath Document Understanding
- ML Skills
- ML Logs
- Document Understanding in AI Center™
- AI Center API
- AI Solutions Templates
- How To
- Basic Troubleshooting Guide
AI Units FAQ
- Why are AI Units consumed even though there
are no attended or unattended robots running?
Pipelines and deployed ML Skills consume AI Units even if no automation is running. Deployed ML Skills hosted on GPU consume a significant amount of AI Units (the default is 40 AI Units per hour, which can add up quickly to about 30,000 AI Units per month). To avoid this, stop the ML Skill when not in use or update your skills to not use GPU. In most situations, we do not recommend using GPU for ML Skills. For more information on changing this, check the Managing ML Skills page.
You can see how much the ML Skill is consuming, by checking the ML Skill Details page.
- Why are AI Robots licenses not displayed even though I am using an Enterprise trial license? Do I also need Document Understanding
pages besides AI Units for production?Since November 2022, AI Units, the new licensing approach for all AI products, has replaced AI Robots. This is visible for the following products:
- AI Center™
- Communications Mining
- Document Understanding
- Task Mining
- How can I track AI Units that are consumed from an Automation Cloud Orchestrator?AI Units are tracked in one of two places:
- Why are AI Robots and Document Understanding pages displayed, even though I bought AI Units licenses? AI Units licenses are
In this situation, your account was probably not fully migrated to the new licensing system yet. In this case, please open a support ticket with UiPath support.
- What happens once AI Units are fully consumed or are expired?
When using Enterprise licenses, once all AI Units are consumed you can't deploy a new ML Skill or start a new pipeline. In order to not break any production flow, you can still keep using an already deployed ML Skill (predictions will be counted as overconsumption) or finish a pipeline that is already started.
- What is the hosting cost in AI Units per hour and prediction?
For AI Units (hourly) you are charged only when the ML Skill is in the Available or Updating status. When the ML Skill is not Available, you will see how much you are charged if the ML Skill is redeployed. See a more detailed AI Unit consumption reporting example in the ML Skills info section.