Azure Open Datasets pricing
Cloud platform to host and share curated open datasets to accelerate development of machine learning models
Azure Open Datasets are a collection of ML-ready datasets from the open and public domain, hosted on Azure. See the overview page to learn more about Azure Open Datasets.
Microsoft pays for the storage costs associated with hosting Azure Open Datasets. While storage will always remain free, egress costs associated with reading large datasets can be charged to the Azure subscription accessing the data Most datasets will be free to access. Egress charges, if applicable, will be mentioned on the Open Datasets overview page.
Azure pricing and purchasing options
Connect with us directly
Get a walkthrough of Azure pricing. Understand pricing for your cloud solution, learn about cost optimisation and request a custom proposal.Talk to a sales specialist
See ways to purchase
Purchase Azure services through the Azure website, a Microsoft representative or an Azure partner.Explore your options
Learn more about Azure Open Datasets features and capabilities.
Estimate your expected monthly costs for using any combination of Azure products.
Review the Service Level Agreement for Azure Open Datasets.
Review technical tutorials, videos, and more Azure Open Datasets resources.
Frequently asked questions
Egress costs mean the cost of reading data from Azure Blob storage, which typically includes read operations and network bandwidth for data leaving the Azure region. See Azure storage pricing page for details.
Your Azure subscription will be charged for compute such as virtual machine instances and any other services such as Azure Machine Learning used to work with Azure Open Datasets.
Datasets are stored on Azure Blob Storage. The storage location is specific to each dataset and is mentioned on the dataset overview page. Allocating compute resources in the same region is recommended for low latency access to the data.
Microsoft has no control over the quality or availability of the original source of data. Hence there is no SLA or availability guarantee for the datasets.