Power BI Desktop builds reports and semantic models on Windows, while Power BI Service hosts, refreshes, and shares them in the cloud.
| Feature | Power BI Desktop | Power BI Service |
|---|---|---|
| Platform | Windows application | Cloud-based web platform |
| Primary purpose | Report and data model development | Report management, collaboration, sharing |
| Installation | Required | Browser access |
| Data connectivity | Broad connector support | Supported cloud sources and on-premises sources through a data gateway |
| Data transformation | Full desktop Power Query experience | Web Power Query with limitations |
| Data modeling | Comprehensive authoring capabilities | Browser-based modeling with limitations |
| Report creation | Yes | Yes |
| Dashboards | Cannot create Power BI dashboards | Can create dashboards |
| Data refresh | Manual and development-time | Scheduled and on-demand |
| Sharing | Publish to Service | Share reports, dashboards, and apps |
| Collaboration | Primarily local development | Shared workspaces and permissions |
| Licensing | Free application | Free, Pro, PPU, and capacity-dependent features |
Power BI Desktop is a free Windows application where analysts connect to data, clean it with Power Query, and design report pages.
Analysts also use it to define the semantic model behind each report. The model contains tables and relationships, and analysts add calculated columns or DAX measures as needed. A .pbix file packages the report and model together, while a Power BI project (.pbip) stores their definitions in separate folders as text files suitable for source control.
Once a report is ready, the Publish command uploads it and its semantic model to a workspace in Power BI Service, where teams can refresh, share, and manage the content. Publishing copies the content, so the local file and the published report stay separate until the analyst publishes again.
Power BI Service, also called Power BI Online, is the browser-based part of Power BI that hosts published content. Organizations use it to store reports and semantic models, schedule refreshes, control access, and distribute content to viewers. Workspaces hold the published content, while dashboards collect tiles from reports or other sources on a single canvas. Power BI apps package selected workspace content for wider distribution.
Power BI Service is part of Microsoft Fabric, which also includes services for working with lakehouses and data pipelines. Organizations can use Power BI without buying dedicated Fabric capacity by working in My workspace or a shared workspace covered by Pro or Premium Per User licensing. Free-license users can view shared content on qualifying Power BI Premium capacity or Fabric F64-or-higher capacity, provided they have the required Viewer access.
The two overlap on report creation but support different stages of analytics work. Desktop gives analysts a local development environment, while the Service handles online distribution, shared access, and administration.
Desktop offers the full Power Query editor and a broad list of connectors, including sources that rely on local files or components. The Service supports compatible cloud sources and on-premises sources through a data gateway, but connector and refresh support can differ between environments. Some connectors available in Desktop don't support refresh in the Service, so analysts need to check compatibility before publishing.
Power Query Online supports transformations through dataflows and the browser-based semantic model editor, although its capabilities differ from the Desktop editor. For sources behind a firewall or on a private network, an administrator or data owner must configure a gateway before the Service can access them for applicable operations.
Desktop remains the best place for complex modeling, with model view, DAX query view, and support for external tools. Analysts can revise tables, relationships, and measures locally before publishing the model.
The Service also supports editing an existing semantic model in the browser. Users with the required write permissions can change relationships, create measures and calculated columns, and define row-level security roles when the model supports those operations. The web editor doesn't support some models and features, and tenant settings can restrict access.
Both environments support creating and editing reports. Desktop provides a local report canvas, while the Service lets authorized users edit reports in a browser. A Power BI report uses a semantic model and can contain multiple pages of interactive visuals and filters.
A dashboard is a separate Power BI item that combines tiles on one canvas. Users create dashboards in the Service by pinning visuals from published reports or other supported sources; Desktop doesn't provide dashboard authoring.
Selecting Refresh in Desktop reloads imported data into the local file. Publishing then uploads the model with the data available at that time; the Service doesn't automatically repeat the local refresh. Analysts configure scheduled or on-demand refresh on the published semantic model and provide the required credentials. On-premises sources also need a suitable data gateway.
DirectQuery models query the source rather than storing a complete imported copy. The Service doesn't run a scheduled source-data refresh for DirectQuery models, although it can refresh dashboard tiles and cached content.
A .pbix file on a laptop remains a local working copy. Colleagues can't open a local .pbix file in a browser, so they need either a published report or a copy of the file. A published report lives in a workspace, where permissions determine who can view or edit it.
The analyst can share an individual report, send an appropriate sharing link, or distribute content through a Power BI app. Publishing alone doesn't grant every viewer access. Each person also needs the correct permissions and a suitable license, unless the workspace runs on qualifying capacity that allows free-license viewing.
In Desktop, analysts define row-level security roles using DAX filter expressions and test them with the View as option. After publishing, an owner assigns users or groups to those roles in the semantic model's security settings in the Service, which also offers a Test as role option.
Report permissions control who can open the report, while row-level security controls which data rows a user can see. Row-level security applies to users with the Viewer role, but it doesn't restrict users with the Admin, Member, or Contributor role, who can edit content.
Consider an analyst who develops a monthly sales report in Desktop and publishes it for the sales team. The analyst prepares the data and report locally, then uses the Service to manage access, refreshes, and distribution.
Desktop costs nothing, so licensing questions begin when content leaves the analyst's computer. In the Service, user licenses determine who can create and share Power BI content, while qualifying capacity can remove the paid-license requirement for eligible viewers.
Yes, Microsoft offers Power BI Desktop at no charge, and you don't need a Pro or Premium Per User license to build reports locally. Sharing is a separate matter because publishing to a shared workspace and distributing results to colleagues depends on Service licensing. A .pbix file can travel by email to another Desktop user, but it carries the loaded data and sits outside workspace permissions and row-level security.
No, a free license lets you sign in to the Service and work in My workspace, but it doesn't let you share content with other users. Pro or Premium Per User licenses generally cover publishing and sharing in standard shared workspaces; PPU also grants access to additional Power BI Premium features, subject to its licensing rules. When a workspace sits on Fabric F64 or larger capacity or qualifying Power BI Premium capacity, eligible free-license users with Viewer access can view its content. However, creating and sharing Power BI content still requires a suitable paid license.
Yes, internal viewers can open reports, dashboards, and apps with a free license when the content sits in a workspace on Fabric F64 or larger, or qualifying Power BI Premium capacity, and they hold Viewer access. Smaller Fabric SKUs don't qualify for this free-viewing arrangement, and tenant settings still govern access. External guests invited through Microsoft Entra B2B require the correct external-sharing permissions and licensing; customer-facing embedded analytics follows a different model, where the application can authenticate its users without individual Power BI licenses, provided the publisher uses suitable capacity and configuration.
Desktop fits local report development and complex model work, while the Service fits browser-based editing, sharing, scheduled refresh, dashboards, and workspace administration. Many reporting workflows use both, as the table below shows.
| Your requirement | Recommended tool |
|---|---|
| Develop complex data models | Power BI Desktop |
| Create a report locally | Power BI Desktop |
| Create or edit a report in a browser | Power BI Service |
| Publish reports to coworkers | Power BI Service |
| Configure scheduled refresh | Power BI Service |
| Create Power BI dashboards | Power BI Service |
| Manage workspace permissions | Power BI Service |
| Build and distribute enterprise reports | Both |
Browser editing depends on the workspace role, model compatibility, and tenant settings. Publishing and sharing also depend on the licensing requirements described above.
Each environment has its own constraints. Desktop runs only on Windows, so Mac users need a Windows virtual machine, a cloud PC, or another supported way to access it. The Service supports browser-based authoring, but its modeling and connector capabilities don't cover every task available in Desktop.
Desktop suits local report development and complex semantic model work. Power BI Service takes over when teams need online access, shared reports, dashboards, refresh schedules, and permission management. Browser-based authoring can be enough for reports that build on existing semantic models, but it doesn't replace every Desktop development capability. For many organizations, the practical choice is to use Desktop to build the report and the Service to publish, maintain, and distribute it.
Yes, Power BI Desktop can connect to data, build models, and save reports as local files. Analysts must manage local data refresh themselves, and they can't use the Service's online sharing or workspace features without publishing content there.
Yes, you can create reports in the browser from existing semantic models, uploaded files, or supported cloud data sources without installing Desktop. Complex modeling and some Power Query work are still easier in Desktop, and browser authoring depends on your workspace role and tenant settings.
Yes, after publication, users with edit access to the workspace can change report pages and visuals directly in the browser. Those edits exist only in the Service copy, so republishing an older .pbix file can overwrite them.
No, Power BI Service supports browser-based report editing and some model changes, but Desktop still offers more complete development tools. Users can build reports on an existing semantic model entirely in the browser.
No, Power BI Service doesn't automatically refresh imported data, so users must configure scheduled or on-demand semantic model refresh in its settings. DirectQuery reports query their source during report interactions, but unavailable sources, credential problems, or gateway issues can still interrupt data access.
Yes, users can usually download a published report as a .pbix file and open it in Power BI Desktop for further editing. Downloading requires sufficient workspace permissions, and tenant settings, report type, semantic model configuration, or size limits can make the option unavailable.
No, Power BI Service is the cloud platform, while Power BI Pro is a per-user license for publishing and sharing content. Free and Premium Per User licenses also work in the Service, and qualifying Fabric capacity changes who needs a paid license.