Guidance Needed on Publishing a Databricks App to the Databricks Marketplace

[pragya17s]

Hi everyone,

We are a Databricks Partner now and Salesforce Partner with over 15 years of experience. We have developed a Databricks App using Free Edition and seek for publication of our App on the Databricks Marketplace. We would appreciate guidance from anyone who has experience with the Marketplace publishing process, particularly for Databricks Apps (Streamlit).

We have a few questions regarding the overall process and would be grateful for any insights or best practices.

Workspace Requirements

• Does Databricks provide a partner sandbox or development workspace for Databricks Partners building Marketplace applications? And if yes , Does it include the necessary functionality to develop, test, and publish Databricks Apps to the Marketplace?
• If we develop and submit our application using the Databricks free trial that includes $400 in credits for 14 days, what happens after the trial expires?
• Or Is a paid Databricks workspace mandatory for developing and publishing Marketplace Apps?
• What is the recommended long-term deployment model for Marketplace applications?

Also our application provides interactive dashboards and AI-powered analytics . We are not publishing a Delta Sharing dataset or acting as a data provider , so do we need Data Provider Partnership, Delta Sharing validation, and preparing sample datasets as onboarding steps or if there is a separate onboarding and publication process for application providers?

Thank you in advance for your guidance and support!

Pragya Sharma ,

Astrea IT Services( pragya@astreait.com )

[Ashwin_DSA]

Hi @pragya17s,

If you want to publish through Databricks Marketplace, you should plan around a paid Databricks environment rather than Free Edition. The Marketplace provider requirements call for a Databricks account on Premium plan or above plus a Unity Catalog-enabled workspace. Separately, Databricks Free Edition limitations explicitly state that Free Edition accounts are for non-commercial use and cannot become Databricks Marketplace providers.

For Databricks Apps specifically, the public docs also say that Databricks Apps supports frameworks such as Streamlit, Dash, and Gradio, and that apps run on Databricks serverless infrastructure. If you are on Azure Databricks, the product docs are even more explicit that Databricks Apps require a Premium tier workspace.

So... I would not treat either Free Edition or a short-lived free trial as the long-term base for Marketplace publishing. The free trial is useful for evaluation, and you can upgrade during the trial by adding a payment method. If you add a payment method or sign up through AWS Marketplace, the account can convert to pay-as-you-go after trial credits are exhausted. But for a real Marketplace publishing path, the safest assumption is to still use a paid Premium workspace.

On the app publication model itself...Marketplace now explicitly supports apps, and third-party apps are installed into the customer's Databricks environment. The docs say that after a vendor submits an app for upload, it goes through a comprehensive security review before it can be published. They also note that Marketplace apps run within the customer's Databricks environment and use the customer’s compute resources, so customer data stays in their workspace. See Get access to third-party apps. The June 2026 release notes also confirm that third-party apps can now be discovered, installed, and run from Marketplace, and that the provider’s code stays closed-source to the customer.

That is also why the long-term deployment model for Marketplace apps is different from a classic data-sharing listing. You build and validate the app in your own workspace, but the installed app ultimately runs in the customer's Unity Catalog-governed Databricks workspace using resources such as jobs, SQL warehouses, Unity Catalog assets, and app compute settings chosen during install.

On the onboarding question...if your offering is an app rather than a Delta Sharing dataset, then the dataset-specific steps should not be the main path. Marketplace supports apps as a first-class listing category, and consumers install them through the Apps flow rather than by receiving a shared catalog. That said, the public provider setup docs still point public Marketplace providers to the general provider onboarding path in the Databricks Data Partner Program. So my reading is...app providers still need the Marketplace provider onboarding path, but they should not assume they must follow every dataset-oriented step such as Delta Sharing validation unless their app also includes a data-sharing component.

I have not seen public documentation describing a Databricks-provided partner sandbox specifically for Marketplace app publishing, so I would plan on using your own paid Premium workspace for development, testing, and provider onboarding.

Hope this helps.

If this answer resolves your question, could you mark it as “Accept as Solution”? That helps other users quickly find the correct fix.

[iyashk-DB]

Yes, Databricks Marketplace now supports third-party Databricks Apps in Public Preview, so partner-built Streamlit/data and AI apps can be published for customers to discover and install in their own Databricks workspaces.

For publishing, the documented baseline requirement is a Databricks account on Premium (or above) and at least one Unity Catalog-enabled workspace, so a free trial / Free Edition may be useful for initial prototyping, but it should not be treated as the long-term workspace for Marketplace publishing and lifecycle management.

Databricks does provide partner build resources through the Partner Program, including sandbox/test-demo environments and Partner Portal access for technical enablement.

For Databricks Apps specifically, the long-term model is that customers install and run the app natively inside their own Databricks environment, using their compute, while the provider code remains closed-source by default.

Since your offer is an application rather than a Delta Sharing dataset, the main path should be app-provider onboarding plus security review, not the standard dataset/data-provider flow unless you are also publishing shared data assets as part of the offer.