Power BI September 2026 Feature Summary

Welcome to the September 2026 Power BI Feature Summary!

As the Power BI, Microsoft Fabric, data, and analytics community gathers in Barcelona for the European Microsoft Fabric + SQL Community Conference, this month is packed with news. From new Copilot and AI experiences to reporting, modeling, data connectivity, developer enhancements, and mobile updates, we're sharing the latest innovations designed to help you build, analyze, and act on your data more effectively. Whether you're joining us in person or following along from afar, this roundup is your guide to everything new across Power BI this month. 

Copilot and AI 

Power BI Agentic Experiences (Generally Available)

Agentic development in Power BI is real, and it is already delivering value: whether you are starting from scratch with greenfield projects or building upon existing solutions in brownfield scenarios, BI developers can delegate the work to an agent, review the result, and refine it rather than starting from scratch every time.

What makes this work is tools and context. Through Work IQ and Fabric IQ, the agent grounds its work in your business context—your data, your semantics, and the way your organization operates. Through skills—published by Microsoft in skills-for-fabric, shared by the community, or personalized by your own team—it also picks up your development context: the modeling and reporting practices, conventions, and workflows your team expects.

Through tools such as the Power BI Authoring MCP server and the Power BI Desktop Bridge CLI, the agent can act on that context: reading and changing semantic models and reports directly, reloading its edits in Power BI tools, and reviewing its own work. Point an agent at a requirements document or a screenshot of the layout you want, describe what you need, and the agent can help accelerate the initial authoring pass, which you can then review and refine.

Figure: Animated GIF showing the creation of a Power BI report based on a remote semantic model using Modeling MCP, the Modeling skill, and report authoring skills.

What's available today

• The powerbi-authoring plugin in microsoft/skills-for-fabric bundling the semantic modeling and reporting agent skills and tools. It was recently updated to support report authoring for remote semantic models in Power BI service and an expanded range of supported report visuals, including custom visuals.
• To learn more, refer to the Power BI skills documentation.
• The local Power BI Authoring MCP server allows the agent to edit semantic models in Power BI Desktop.
• To learn more, refer to the Power BI Authoring MCP documentation.
• Power BI Desktop Bridge lets an agent reload changes in a running instance of Power BI Desktop and capture screenshots to check its own work.
• To learn more, refer to the Power BI Desktop Bridge documentation.
• Browser-based validation for remote report editing, so the agent can validate a report using a headless browser (an automated browser that runs without a visible user interface). Removing the local Power BI Desktop dependency allows the full authoring loop to run without Power BI Desktop, extending the agentic experience to Mac users and others who do not rely on a local installation.

Microsoft 365 Copilot is coming to Power BI (Preview)

Copilot experiences have emerged across the applications people use every day, bringing AI closer to the work happening in each product. But this can also feel fragmented, with different entry points, separate conversations, and uncertainty about which Copilot agent to use.

Our goal is to create a consistent Copilot experience that is available wherever people work. People should not have to leave the application they are using to find it. Microsoft 365 Copilot should be available within their experience, carry relevant context across experiences, and connect to the specialized intelligence of each product.

We’re taking an important step toward that goal with a preview of Microsoft 365 Copilot integrated directly into Power BI and Microsoft Fabric.

This integration gives users a familiar place to bring together business context and trusted data. Shared conversation history lets them continue a discussion as they move between Microsoft 365 Copilot experiences instead of starting over in each application.

Figure:Screenshot of Copilot Chat integrated into the Power BI experience, including integration with Work IQ.

In Power BI, a question no longer has to stop with the numbers in a report. A sales leader could use Fabric IQ to identify which regions are behind plan, then use Work IQ to connect the results with customer emails, meetings, account plans, and documents. They could summarize the complete picture and even kick off actions and tasks in Cowork, all from inside Power BI. This work context and ability to move directly from insight to action were not available in the standalone Power BI Copilot experience.

Figure:Screenshot of the Cowork integration into the Power BI experience, highlighting the full Microsoft 365 experience embedded into Power BI.

A new setting in the Fabric admin portal will determine whether this integration is available by enabling or disabling its entry point in the navigation pane. It will be off by default, giving administrators time to evaluate the experience before introducing it to their organization. Administrators can enable it when they are ready and disable it if they need to pause access.

This setting is independent of the existing setting for standalone Power BI Copilot. If both settings are enabled, users with a Microsoft 365 Copilot license will see entry points for both experiences, and standalone Power BI Copilot will continue to appear on Home. If standalone Power BI Copilot is disabled and Microsoft 365 Copilot is enabled, users will see only the Microsoft 365 Copilot entry point.

During the initial preview, the integrated experience is available only to users with a Microsoft 365 Copilot license. These users receive the full Copilot Chat experience in Power BI and Fabric, bringing together Fabric data through Fabric IQ and relevant Microsoft 365 context through Work IQ. Existing permissions and governance controls continue to determine which content each user can access.

Support for a broader Fabric audience will follow. This expansion will give Fabric users without a Microsoft 365 Copilot license access to an integrated Copilot experience tailored to their available capabilities. The experience may differ from the full licensed offering, but it will extend Fabric IQ and conversational data experiences to more types of Fabric users.

The preview will roll out with the setting turned off by default, with administrators choosing when to enable it for their organization.

This is an important step toward one Copilot that meets people wherever they work, connecting the context surrounding each decision with trusted business data.

Power BI data answers in Microsoft Copilot Chat (Generally Available)

Users can find and ask questions about Power BI reports and semantic models directly from Microsoft 365 Copilot. They can search for relevant Power BI content, attach a report or provide a link, and ask questions in natural language.

Figure: Microsoft Copilot Chat featuring a data answer coming from a Power BI report, cited.

Powered by Fabric IQ, the experience uses the data, measures, relationships, and business logic already defined in Power BI to provide grounded answers. Users can also bring Power BI insights into conversations that include their Microsoft 365 work context, such as emails, meetings, and documents.

Microsoft 365 Copilot now supports Microsoft Purview Data Loss Prevention (DLP) policies and information protection controls for this experience. This helps organizations adopt conversational analytics while applying their existing policies for identifying, classifying, and protecting sensitive data. The experience continues to respect existing Power BI permissions and Microsoft 365 governance controls. Users can only access Power BI content they are authorized to view, and administrators retain control over whether Fabric data is available in Microsoft 365 Copilot.

Organizations can use Power BI data answering in Microsoft 365 Copilot with production support and service commitments. This milestone makes it easier for users to bring trusted business data into the flow of work without leaving Microsoft 365 Copilot or creating a separate agent.

Fabric IQ Plugin in Microsoft 365 Copilot Cowork (Generally Available)

Power BI data is available where multi-step work happens, so a business question, the analysis behind it, and the follow-up it triggers stay in one place.

To ground a task in Power BI, users can search for relevant content, attach a report, or paste a link to a report or semantic model. Questions are answered using the data, measures, relationships, and business logic already defined in the semantic model. Coverage now extends to reports published through organizational apps, with improvements to content discovery, large model handling, schema selection, Verified Answers, and row-level value search.

The plugin also carries forward the governance controls organizations rely on in Power BI:

• Existing access controls: Power BI permissions and row-level security determine which reports and semantic models each user can reach.
• Citations and sensitivity labels: Responses cite the Power BI content behind an answer and surface the Microsoft Purview Information Protection label applied to it, so users can see where a number came from and how it is classified.
• From insight to action: With Fabric IQ, users can analyze Power BI reports using natural language, uncover key findings, and put those findings to work right away in Cowork. Whether they are summarizing business performance, investigating a trend, or answering a data question, Fabric IQ turns trusted organizational data into outcomes.

Build Fabric Apps more easily in GitHub Copilot App (Preview)

Fabric Apps were introduced in June as an AI-first way to build custom web apps with Microsoft Fabric as the backend. This month we’re excited to share that building these Fabric Apps on your governed data has become even easier with the GitHub Copilot App, which gives you a natural language chat interface instead of a command-line interface.

Figure: GitHub Copilot App chat interface.

To try it out, start by downloading the GitHub Copilot App. Then navigate to Customize > Plugins > Rayfin, and select Install to enable our dedicated plugin:

Figure: Installing Rayfin plugin in GitHub Copilot App.

Once the plugin is enabled, simply start a new chat and work with Copilot to help plan your Fabric data app. After you review and approve the plan, ask the agent to execute it and specify the workspace where you want to deploy the app.

Figure: Example New York Taxi app.Enhanced Report Consumption Copilot with Fabric IQ

When viewing a Power BI report, you can enable Fabric IQ from the Copilot pane to better answer complex questions about your data. This enhancement merges existing Fabric IQ capabilities into Power BI Copilot report viewing mode, allowing Copilot to use a multi-step approach to evaluate data in the semantic model and refine its approach when generating a response.

Figure: Fabric IQ enabled in Power BI Copilot report pane (viewing mode) to answer a data question with a supporting textual answer and visualization.

With Fabric IQ enabled, Copilot can:

• Provide better answers to questions about your data based on detailed reasoning capabilities directly into the report viewing experience.
• Show reasoning steps to help you understand how Copilot provided an answer.
• Generate Fabric visuals to help you further analyze your data. Answers can include column, bar, stacked, line, area, pie, donut, scatter visuals.
• Provide the underlying DAX query and data table used to generate a response.

Figure: GIF illustrating how to enable Fabric IQ in Power BI Copilot report pane (viewing mode) to answer data questions.New control for Copilot access to semantic modelsPower BI now provides semantic model editors with greater control over how Copilot can be used with their data. Users with write permissions can block read-only users from accessing a semantic model through Copilot and Microsoft’s remote MCP tools, supporting governance, compliance, and access-control requirements.

The setting applies across Copilot experiences that query the semantic model, including Power BI Copilot experiences; Microsoft 365 Copilot Chat and Cowork; and data agents, remote Fabric MCP servers, and Power BI MCP servers. When disabled, users with read-only access won't be able to access that semantic model through these experiences. By default, the setting remains enabled to preserve the current behavior.

This setting appears in the Copilot section with the following description: “Allow any person with read-only access to use AI (i.e. Copilot, Microsoft’s MCP tools) on the model and its related reports.”

Figure: Setting that allows people with read-only permission to use AI experiences with the semantic model and its related reports.

This update gives organizations a clear and enforceable way to control AI access to semantic model data while maintaining flexibility for semantic model editors.

To learn more, refer to the Use Copilot with Semantic Models in Power BI documentation.

Remote Power BI Authoring MCP Server (Preview)

The Power BI Authoring MCP enables AI applications to create and modify semantic models in Fabric workspaces, without installing or maintaining anything local. This release also unblocks Mac users that want to use the Power BI Authoring MCP, because the server runs remotely and does not require local binaries.

Use the following endpoint in AI applications such as GitHub Copilot App:

https://api.fabric.microsoft.com/v1/mcp/powerbi/authoring

Figure: Power BI Remote MCP Server configuration in GitHub Copilot App.

The local Power BI Authoring MCP server, released in November 2025, remains generally available. It remains an option when an agent needs to work with Power BI Desktop, Power BI Project (PBIP) or TMDL files on disk. The local and remote servers support the same authoring capabilities.

We recommend using the remote server when local development is not required, such as when creating a Direct Lake semantic model in a Fabric workspace.

In the following example, the AI agent uses one tool call to create a Direct Lake semantic model connected to a Fabric Lakehouse:

Figure: Using Power BI Remote MCP to create a Direct Lake semantic model.

Reporting 

New Report Ribbon on the web (Preview)

The Report ribbon is now available on the web, providing a consistent experience for report consumption and authoring that more closely matches Power BI Desktop. The ribbon organizes commands into clear, easy-to-navigate tabs, making it easier for users to discover features, interact with reports, and complete editing and authoring tasks directly in the browser. It also introduces the new Design ribbon and additional capabilities that were previously unavailable on the web, such as Format Painter.

This milestone brings greater consistency to Power BI report-authoring experiences and marks an important step toward a seamless, end-to-end reporting workflow on the web. As web authoring capabilities continue to expand, users will be able to accomplish more of their reporting tasks directly in the browser, enabling a more streamlined, productive workflow.

Figure: New report ribbon on the web.

Enable the Try the new ribbon (preview) at the top of the page to explore the new experience. You can disable the setting to return to the previous ribbon.

Live Visual Sharing in Outlook on the web (Preview)

For years, sharing data through email has been stagnant. You could send a screenshot, but it immediately became stale. You could attach a report link, but recipients had to leave their workflow to find the context they needed. And every follow-up question often meant another screenshot, another export, or another email thread.

Now, we're bringing a new way to share Power BI insights directly in the flow of work with Live Visual Sharing in Outlook. With Live Visual Sharing, you can insert a Power BI visual into Outlook on the web and share it as a live, interactive experience. Recipients can view and interact with the visual directly in Outlook while staying connected to the latest underlying data.

Figure: A live Power BI visual in Outlook on the web.

When a live visual is shared, Outlook renders an interactive version of the Power BI visual. Rather than embedding a static image, the experience remains connected to the source report and retrieves current data when viewed.

Recipients can:

• Interact with the visual directly in Outlook on the web.
• Access tooltips and explore data points.
• See filters and slicers applied at the time of sharing.
• View refresh information and data context.
• Navigate back to the source report when deeper analysis is needed.

The result is a richer conversation around data, directly within Outlook.

Sharing a live visual is simple:

1. Open a Power BI report in the service.
2. Select a visual to share, then select More options in the upper-right corner.
3. Select Share.

Figure: Share option in Power BI.

4. Select Link to this visual.

Figure: Link to this visual as part of the share dialog.

5. A pop-up will appear asking if you want to use the current view (apply existing filters and slicers) or not.
6. After selecting Create Link, the link is copied to your clipboard.

Figure: Before copying the link, you can configure the view in this dialog.

7. Open Outlook on the web. This experience is not available in the Outlook desktop app. Create a new email.
8. Paste the link in the email. The visual renders in loop after a few seconds! You can also share the same link in a Loop page, and the visual will also be rendered.

Figure: An unfurled Power BI visual in Outlook on the web.

By combining Power BI, Outlook, and Loop-powered experiences, we're bringing data closer to where collaboration happens. Live Visual Sharing can support more informed, data-driven conversations in Outlook.

To learn more about Live Visual Sharing in Outlook, refer to the Share live Power BI visuals documentation.

More control and context in Power BI Home (Generally Available)

Power BI Home now provides additional context and personalization options to help users discover the right content more efficiently. With new freshness indicators and customizable list views, you can quickly identify the most relevant reports and semantic models, prioritize the metadata that matters most, and tailor the experience to fit your workflow.

Find the right content more easily

Discover the right content more easily with new Data Refreshed and Modified Date columns in Power BI Home. By bringing the freshness signals already familiar from workspace list views into Home, users can more easily spot recently updated reports and semantic models, distinguish between similar items, and quickly find the content that matters most. Whether you're checking for the latest business updates or deciding which refreshed report to open, you can navigate with greater confidence and less guesswork.

Figure: Power BI Home showing two new columns for Data refreshed and Modified date.

Customize your Home list view

Not every user, team, or organization relies on the same signals to discover content. With Column Options, you can show or hide metadata, drag and reorder columns to prioritize the information you use most, and reset to the default view at any time. Whether you rely on freshness signals, endorsements, sensitivity labels, or location information, you can tailor Home to match your workflow.

Figure: Power BI Home showing new column options dropdown.

Set and reset column widths in table and matrix (Generally Available)

You can now right-click a column header in a table or matrix and select Set column width to specify its width in pixels. This option helps report viewers improve readability and fit the information they need within the available space, without changing the report author’s design or affecting other viewers. Changes made by viewers last only for the current viewing session. To restore the width defined by the report author, right-click the column header and select Reset column width.

Figure: Set or reset the width of an individual table or matrix column from its context menu.

Visual calculations for small multiples (Generally Available)

You can now use visual calculations in visuals with small multiples. Create calculations such as moving averages and compare trends across categories without adding separate measures to the semantic model. In the visual matrix, the field in the small multiples field well appears on the ROWPAGES axis. You can select ROWPAGES by itself or combine them with ROWS and COLUMNS to control how the calculation moves through the visual.

Figure: Create a moving average visual calculation for a line chart divided into small multiples by country.

Design ribbon for Power BI reports (Generally Available)

The new Design ribbon brings report styling tools together in one location. Apply a built-in or organizational theme, open the Theme pane to customize report-wide defaults, or import and download theme files directly from the ribbon. You can also quickly open Page size, Page background, and Wallpaper settings for the current page.

Figure: The Design ribbon provides theme controls and shortcuts to page formatting settings in Power BI Desktop.

Two new built-in themes, Dark and Soft dark, use the new Versa color palette and styling that closely follows the Fluent 2 base theme. These themes let you design against a dark report canvas and return to the default light theme before publishing if needed.

Figure: Apply the Dark built-in theme from the Themes gallery on the Design ribbon.

Slider color options for Slicer visual (Generally Available)

Match your slicer's slider to your report design. The Slicer visual's Between, Before, After, and Date picker settings include a Slider format section where you can turn the slider on or off, set the handle style, and choose colors for the handles, selected range, and track.

Handle options adjust based on the Responsive layout toggle (General > Properties > Advanced options). When Responsive is on, handles appear as circles with Fill color, Border, and Size.

Figure: Slider formatting options for a date range slicer with Responsive layout turned on.

When Responsive is off, handles appear as rectangular bars with Border, Width, and Height.

Figure: Slider formatting options for a date range slicer with Responsive layout turned off.

Translytical task flows show user date function (UDF) information in button action (Generally Available)

The formatting pane now displays the Workspace, Function set, and Data function assigned to a data function button. These read-only details help you confirm that the button references the intended function without reopening the OneLake catalog. This information is especially useful when you review or troubleshoot reports that use several functions or connect to functions across multiple workspaces.

Figure: View the workspace, function set, and data function assigned to a button in the formatting pane.

Format pane headers stay visible as you scroll (Generally Available)

Expanded section headers now remain visible at the top of the Format visual pane as you scroll through their settings. This update helps you keep track of the section you’re editing, especially when a section contains many formatting options. For example, when you scroll through color settings, the Colors and Data colors headers remain visible for context.

Figure: Expanded section headers remain visible while you scroll through formatting settings.

Pick the axis initial scroll position (Generally Available)

Categorical axes—like month, year, or product name—show a scrollbar when there are more categories than fit in the visual. Until now, the scrollbar always started at the beginning of the data, so the oldest values displayed first. This month, you can set Initial scroll position to End on the axis Layout section, and the visual loads with the most recent data in view. Users can still scroll back to see earlier history. The option is available on stacked, clustered, and 100% stacked bar and column charts; line, area, stacked area, and 100% stacked area charts; line and stacked column and line and clustered column charts; ribbon, waterfall, and scatter charts with a categorical X-axis.

Figure: Line chart withInitial scroll positionset toEndon the X-axis. The scrollbar is positioned on the right, showing the most recent months first.

Azure Maps geocoding improvements (Generally Available)

Large reference layers now render correctly when you filter your report. Reference layers add geographic context using points, lines, and polygons, such as locations, routes, postal codes, sales territories, and administrative regions. The Azure Maps visual renders up to 30,000 reference-layer features at a time. Previously, when a reference layer exceeded this limit, features excluded from the initial view could remain missing even after a report filter narrowed the data to a smaller area.

Figure: Azure Maps visual displaying an all-up view of a US ZIP Code reference layer containing more than 30,000 polygons. Some ZIP Code boundaries are not displayed because the reference layer exceeds the rendering limit.Figure: Azure Maps visual showing the previous behavior after the report is filtered to Washington. Only ZIP Code polygons included in the initial feature set are displayed, resulting in an incomplete view of the state.Figure: Azure Maps visual showing the new behavior after the report is filtered to Washington. The full reference layer is re-evaluated, displaying all relevant ZIP Code boundaries for the state.

Now, Azure Maps re-evaluates the reference-layer features whenever report filters are applied. The unfiltered view is still limited to 30,000 features, but a filtered view accurately displays the relevant features, up to that limit.

On-object interaction improvements (Preview)

Based on your feedback during preview, we're introducing the first in a series of on-object interaction improvements, designed to make quick edits to your visuals easier. This month we're refining how you add data and manage some visual elements.

The Build flyout now opens only when you select Add data, helping you stay focused when dragging and dropping data directly onto a visual or into the Build pane.

We’ve also separated Title and Subtitle in the Add to your visual flyout, so you can manage them independently. In addition, title and subtitle are now turned off by default for matrix, table, and card visuals in reports that use the latest base theme. New reports automatically use the latest base theme. For existing reports, select Update theme in the Theme pane to apply this change. You can turn either element back on when needed.

Figure: On the left, select Add data to open the Build flyout. On the right, manage Title and Subtitle independently from the Add to your visual menu.

Support for Fabric Apps in Org apps (Coming Soon)

Soon, users will be able to include generative Fabric Apps alongside the reports, dashboards, notebooks, and other content in your Org Apps.

For example, imagine you want people to rate the content in an Org App and share feedback about what would make it more useful. You can create a Fabric App for that experience and add it directly to the Org App.

Figure: Adding a Fabric App item for rating an experience into an Org App and then viewing that rating experience from within the Org App.

Whether you’re helping people explore what could happen next, act on an insight, or complete a specialized workflow, Fabric Apps give you a flexible new way to build for your audience without sending them somewhere else. Together, organizational apps and Fabric Apps provide a way to bring enterprise reporting, content, and tailored experiences together for your organization.

Changes to Power BI subscription email delivery

Power BI is updating the platform used to deliver subscription emails as part of ongoing improvements to delivery reliability and long-term service health.

As eligible subscriptions move to the updated platform, customers may notice changes to the sender identity and email formatting. Each subscription transfers automatically when it moves. Customers do not need to recreate it or change its schedule or recipients.

Figure: After selecting Subscribe for an item, an information icon identifies a subscription that has moved to the updated email delivery platform.

Increased Export to Excel row limits (Generally Available)

Many users export Power BI data to Excel for further analysis. This month, we're increasing the maximum number of rows that can be exported from table visuals using Data with current layout from 150,000 to 500,000. This increase helps you analyze larger datasets in Excel with fewer export operations.

Figure: Export data from a table visual to Excel using Data with current layout, supporting up to 500,000 rows.

Value selection in Narrative visual

We are replacing the Q&A natural language experience for creating dynamic values in the Narrative visual and text box with a more deterministic UI. This is due to the ongoing deprecation of the Q&A tool, and also based on feedback that it was hard to deterministically pick exact values through the Q&A experience. 

Authors can now create and edit custom values through a direct, deterministic UI: select a column or measure, apply any valid summarization, and add any type of filter.  

Auto-generated values are unchanged, and values previously authored with Q&A remain available as read-only. 

Figure: The new value selection options in the Narrative visual.

Modeling 

Create and edit DirectQuery semantic models in the Power BI service

DirectQuery semantic models can now be created and edited directly in the Power BI service, eliminating the need to download and use Power BI Desktop.

Previously, building models that query data sources in real time meant downloading Desktop, authoring the model, and publishing it to the web. Now, web-based modeling gives your team the freedom to:

• Work from any device: Mac, Linux and Windows users can now author DirectQuery models straight from the browser.
• Collaborate in real time: Multiple team members can work on the same semantic model simultaneously.
• Ditch the desktop loop: Skip the download-edit-republish cycle and update your models instantly in the service.

 Capabilities on the web

Start from scratch: Connect to a supported data source and choose DirectQuery mode.

Figure: Power Query interface displaying data source connection settings, with DirectQuery selected as the data connectivity mode.

Full modeling control: Open existing DirectQuery models to manage relationships, write DAX and tweak table/column properties.

Figure: Model View in Power BI Service showing tables, relationships, and the properties panel for editing a DirectQuery semantic model.

Flexible storage modes: Easily switch table storage modes from DirectQuery to Import or Dual if needed.

Figure: Properties pane that allows user to switch table storage modes from DirectQuery to Import or Dual.

Thanks to this new capability, composite models are also available on the web! Head to the Power BI service and open one of your DirectQuery semantic models or create a brand-new one. We'd love to hear what you think!

Approximate Distinct Count comes to Import and Direct Lake (Preview)

Distinct count is one of the most resource-intensive operations in any analytical engine. As data cardinality grows, so do CPU and memory usage, often slowing down your interactive reports.

APPROXIMATEDISTINCTCOUNT already offered a faster alternative for DirectQuery sources such as Azure SQL, Synapse, BigQuery, Databricks, and Snowflake. This month we're extending support to Import, Direct Lake on OneLake, and Direct Lake on SQL semantic models.

DAX = APPROXIMATEDISTINCTCOUNT(<columnName>)

Figure: Model View in Power BI Service showing tables, relationships, and the properties panel for editing a DirectQuery semantic model.

How it works and when to use it

Instead of scanning every single unique value, APPROXIMATEDISTINCTCOUNT calculates a highly accurate estimate of unique values with a standard error rate of ~1.6%. In return, you get faster query execution times.

• Best for high-cardinality columns: Use it on columns with millions of unique values (such as transaction IDs, user IDs, or timestamps) where an approximate result can reduce the resources required to calculate distinct values.
• Keep standard DISTINCTCOUNT for low-cardinality data: Stick to exact counts for smaller attributes like Country, Category, or Status, where standard calculation is already fast, or for financial metrics where absolute precision is required.

Swap APPROXIMATEDISTINCTCOUNT into your heaviest measures today and let us know how much speed you gain on your models!

Accelerate text searches in your semantic models with String Indexing (Preview)

String Indexing for Power BI semantic models is coming soon! If your reports rely heavily on text fields, such as product names, customer comments, or long descriptions, you’ve likely faced slow performance when users search or filter on large semantic models. String Indexing is designed to improve text-search performance in supported semantic models.

Why String Indexing?

Common text operations like DAX functions (SEARCH(), CONTAINSSTRING()) and advanced text slicers in reports traditionally scan data row-by-row. As datasets grow, these scans slow down query times.

String Indexing builds specialized indexes so the engine can locate matching text more efficiently, which can improve report responsiveness for both Import and Direct Lake semantic models.

Indexing modes

You can configure String Indexing via model metadata to fit your workload needs:

• Auto Mode: Automatically builds the index on the first query execution for a qualifying column. The index does not persist in storage; it remains in memory only while the column is loaded and will be rebuilt on subsequent queries if evicted. Ideal for smaller models.
• Full Mode: Builds and persists the index during scheduled data refreshes. While refresh times and model size may slightly increase, users enjoy consistently fast query performance right out of the box.

Get started

String Indexing requires a semantic model running at 1707 compatibility level. During this initial preview phase, indexing can be configured through TMDL on the web or XMLA metadata. Desktop support and a dedicated UI experience within Power BI are planned for a future release to simplify setup.

Figure: Setting up the “stringIndexingbehavior” property to full in TMDL view in the web.

Check out the String Indexing documentation for setting up instructions, prerequisites, and guidance on choosing the right indexing mode for your models.

We look forward to seeing how this feature helps make text-heavy reporting experiences faster and more responsive!

Introducing Full-Text Search in Power BI Semantic Models (Preview)

Full-Text Indexing for Power BI semantic models is coming soon! While traditional string searches look for exact character matches, Full-Text Indexing enables a more intelligent search experience based on the meaning of words and natural language concepts. This capability powers two new DAX functions, TEXTCONTAINS() and TEXTSIMILARITY(), making it easier than ever to analyze large volumes of textual data directly in your semantic models.

Why Full-Text Indexing?

Business data increasingly contains rich text content such as customer feedback, support tickets, product descriptions, incident summaries, and knowledge articles. Searching this data using traditional string matching often requires exact matches and can miss relevant results due to differences in spelling, word forms, or phrasing.

Full-Text Indexing builds a specialized search index at the column level, allowing semantic models to understand words rather than just character sequences. This enables advanced search experiences such as:

• Finding related words through linguistic analysis (for example, run, running, and runs)
• Matching singular and plural variations automatically
• Ignoring common stop words such as "the" or "and"
• Supporting phrase-based searches
• Tolerating minor spelling mistakes through fuzzy matching
• Enabling semantic similarity search scenarios

Getting Started

FTS Indexing requires a semantic model running at 1708 compatibility level. During this initial preview phase, indexing can be configured through TMDL on the web or XMLA metadata. Desktop support and a dedicated UI experience within Power BI are planned for a future release to simplify setup.

Figure: Setting up the “fullTextIndexingbehavior” property to full in TMDL view in the web.

New DAX functions

Full-Text Indexing is required for two new DAX functions:

• TEXTCONTAINS()
  • Returns TRUE when a text value matches a search expression using full-text search semantics rather than simple substring matching. It supports multiple search modes, including keyword matching, phrase matching, and fuzzy matching for typo tolerance
  • TextContains(<WithinTextColumn>, <SearchText>, [MatchMode])
• Example: "Show me all reviews that mention spa." The review either matches or it doesn't.
  Figure: Example of the new TEXTCONTAINS() DAX function in DAX query view.
  
• TEXTSIMILARITY()
  • Returns a relevance score indicating how well the text in <WithinTextColumn> matches <SearchText>. It enables similarity-based text comparisons, helping identify records that are semantically related even when they do not contain the exact same words.
  • TextSimilarity(<WithinTextColumn>, <SearchText> [, <MatchMode>])
• Example: "Show me the 10 reviews most relevant to 'good breakfast'." Reviews mentioning both terms prominently rank higher than reviews that only partially match.

Figure: Example of the new TEXTSIMILARITY() DAX function in DAX query view.

Learn more about these two new DAX functions (TEXTCONTAINS, TEXTSIMILARITY) in our documentation.

Full-Text Indexing is supported for both Import and Direct Lake models, with Full mode recommended (during Preview) for the best out-of-the-box performance and user experience.

Refer to the Full Text Indexing documentation to learn more about prerequisites, configuration options, and setup instructions.

Data connectivity

New Power Query editor (Preview)

Since May of this year, the modern Get Data experience has been available in preview inside Power BI Desktop; the feedback we've gathered from that rollout has directly shaped what comes next. Your reports of friction points, missing capabilities, and rough edges between the legacy and modern interfaces went straight into the backlog, and this month that work pays off: the full modern Power Query Editor experience moves into preview.

Figure: Screenshot of the new Power Query editor using the light theme.

It's the same trusted engine, wrapped in a completely refreshed editing surface — and we're still closing the remaining gaps with the legacy UI as we go.

Let's explore six key reasons to enable and try this feature.

1. Modern Get Data - The same connector browsing and search experience you already know from Fabric Dataflow Gen2 — now in Desktop. Searchable, categorized, and designed to make connectors easier to find and navigate than the legacy dialog.

• Try it: Connect to a new data source and browse the OneLake catalog.

Figure: Screenshot of the Modern Power Query Get Data experience inside of Power BI Desktop.

2. Dark theme, end to end - No more jarring light-mode panes breaking your flow. The legacy editor only partially supported dark theme. The modern editor applies it consistently across the preview grid, dialogs, and other parts of the editing surface. 

   Figure: Screenshot of the Power Query editor using the Power BI dark theme.
3. Folding visibility - Stop guessing whether your query folds. Step folding indicators mark each step as folding or not, and the new query plan shows exactly how the engine intends to evaluate your query — and where folding breaks down. 

• Try it: Open the query plan on a query with several steps and find where folding stops.

Figure: Screenshot of the Query Plan dialog inside of the new Power Query editor.

4. Diagram view - See how your queries relate to each other visually, instead of reverse-engineering it from names. A gamechanger for files you've inherited or models with long reference/merge chains. 

• Try it: Open a file with referenced queries and trace a merge visually.

Figure: Screenshot of the Diagram View inside of the new Power Query editor experience.

5. New capabilities, now on Desktop - Features that used to live only on the web are now here: rank column, cluster values, schema view, query script view, plus upgrades to transforms you already use daily. 

• Try it: Add a rank column or cluster values on a messy text column.

Figure: Screenshot of the Schema view available in the new Power Query editor.

6. Export a Power Query template - Capture your solution at a point in time and take it with you. Export a template from Desktop and import it straight into a Dataflow Gen2 — queries, steps, and structure intact. 

• Try it: Export a template from a multi-query file and import it into a Dataflow Gen2 to compare.

Figure: Screenshot of the Export template experience in the Power Query editor.

Ready to try it?

Enabling the modern Power Query experience takes just a few seconds — no restart required:

1. Open Power BI Desktop.
2. Locate: File > Options and settings > Options.
3. Under Preview features, check New Power Query experience.
4. Open (or reopen) the Power Query Editor — you're now in the modern experience.

Once you've had a chance to explore this new user interface and its experiences, we want to hear from you. 

Microsoft Outlook Mail connector (Preview)

We're introducing the new Microsoft Outlook Mail connector (Preview) for Power BI Desktop. Built on the Microsoft Graph API, it lets you read your mailbox data — folders, messages, and attachments — using an organizational account. This is the first in a set of new connectors we'll be releasing as a modern alternative to the Exchange Online connector.

The initial preview is scoped to Mail in Power BI Desktop. Additional connectors for Calendar Events, Contacts, and Tasks are on the roadmap, along with Excel and Power BI service / Fabric Dataflows Gen2 support.

Scheduled refresh isn't supported yet — this applies to both the Power BI service and refreshes routed through the on-premises data gateway.

To try it, enable the Microsoft Outlook Mail preview feature under Options in Power BI Desktop, then open Get Data → Online Services → Microsoft Outlook Mail and sign in with your organizational account.

Figure: Microsoft Outlook Mail connector with the Preview badge in the Power BI Desktop Get Data experience.Figure: Microsoft Outlook Mail listed among Online Services in the Power BI Desktop Get Data pane.

More flexible Data Privacy Firewall query patterns in Power BI Desktop (Preview)

Power BI Desktop now allows Data Privacy Firewall partitions that reference other partitions also access data sources when the privacy levels of the involved data sources are compatible. Introduced in the July 2026 release and enabled by default, this preview setting aligns Power BI Desktop with Power Query Online, gateways, and the Power BI service. 

This change supports common query patterns in which one query returns a value that another query uses to build a web request, SQL statement, file path, or other data source request. These patterns can now run without refactoring queries or duplicating logic solely to satisfy the previous partition restriction. 

The Data Privacy Firewall remains active. Power Query continues to evaluate the privacy levels of the involved data sources before combining data, helping authors use more flexible query designs while preserving privacy protections.

Figure: The new Power BI Desktop preview setting allows compatible cross-partition query patterns while the Data Privacy Firewall remains active.

Mobile

Mobile page indicator (Generally Available)

Know immediately which report pages are mobile-ready, without clicking through them. Each page that has a mobile layout shows an icon on its page tab in the footer, so you can spot your mobile coverage instantly. Pages without a mobile layout stay unmarked, making it easy to see what's left to optimize.

The icon is coming soon to both Power BI Desktop and the Power BI service. In Power BI Desktop, it appears in both desktop layout view and mobile layout view, so you can track mobile coverage right where you work. In the Power BI service, it appears in mobile layout view.

Figure: Page tabs marked with an icon to show which pages have a mobile layout.

Contact details in Info drawer (Generally Available)

We’re making it easier to connect with the right people from the Power BI Mobile app. You can now find contact details in the Info drawer for reports, apps, and dashboards, and email the relevant contact directly from your phone.

Figure: Two Power BI Mobile screens showing the Contoso Sales workspace and the report’s Info drawer with a clickable contact name for sending an email.

Developers and APIs 

Power BI Projects (PBIP) (Generally Available)

Since entering preview in 2023, PBIP has changed how developers build and manage Power BI content. It stores report and semantic model definitions as human-readable text files, turning a Power BI solution into a codebase.

Developers can search these files, track changes in Git, review diffs in pull requests, automate updates with scripts and pipelines, and use AI agents to work with models and reports. Changes remain visible and reversible, which makes team development easier and safer.

Recent improvements have made the workflow more practical. These include opening a project directly in Visual Studio Code and detecting external file changes in Power BI Desktop.

Figure: Open Visual Studio Code entry point in Power BI Desktop.

New underlying formats capabilities:

• Tabular Model Definition Language (TMDL) for semantic models.
• Power BI Enhanced Report Format (PBIR) for reports. PBIR is now the default report format for PBIX and PBIP files. New reports use it from the start. When you edit a report that uses the legacy format, Power BI tools automatically convert it to PBIR.

New Purge Data option for Fabric APIs

The new Purge Data option for Fabric APIs gives developers and administrators greater control over programmatic semantic model deployments. This capability allows users to explicitly approve deployments that may require removing existing data, helping unblock scenarios such as schema or data type changes that previously failed validation checks.

Designed with both safety and automation in mind, purgeData keeps data protection safeguards enabled by default while giving developers the flexibility they need to manage complex deployment scenarios at scale.

Other 

Data updated and sensitivity fields of Power BI report in Microsoft Teams preview card (Coming Soon)

Power BI report previews in Microsoft Teams will soon surface the report's last refresh date and sensitivity label. These additional signals will help users quickly assess the freshness, relevance, and classification of a report before opening it.

Figure: Power BI report preview card in Microsoft Teams chat showing report metadata, including the last refresh date and sensitivity label.

Artifact info header for Power BI tabs in Microsoft Teams (Generally Available)

Power BI content in Microsoft Teams tabs now includes the artifact info header, giving users the context they need without leaving their flow of work. Available for reports, apps, organizational apps, dashboards, and scorecards, the header brings key artifact information and actions together in one place.

The new header displays the artifact name and sensitivity label and provides an entry point to artifact metadata. You can open the artifact in the Power BI personal app in Teams, open it in a browser, and refresh the tab.

Tab owners can control their experience with the new Show info header setting on the tab settings page. The setting is 'on’ by default and can be turned off when a more compact tab experience is preferred.

Figure: Power BI report in a Microsoft Teams tab with the artifact header open, displaying report metadata.Figure: Power BI report in a Microsoft Teams tab, with the setting dialog open.

Third-party custom visuals

Drill Down Bubble PRO by ZoomCharts

Drill Down Bubble PRO is a Power BI visual designed for exploring relationships, trends, and outliers across multiple dimensions of data. It adds richer exploration, selection, and analytical capabilities directly inside the visual.

Key features include:

• Multi-level drill down to explore data hierarchies from a high-level overview to detailed insights.
• Pan and zoom controls to navigate dense datasets and focus on specific clusters or individual data points.
• Area and lasso selection to quickly isolate groups of bubbles and use selections for interactive analysis and filtering.
• Dynamic regression lines that recalculate as the displayed data changes, making trends and relationships easier to identify.
• Line and area thresholds to highlight targets, ranges, exceptions, or other important business benchmarks.
• Area shading to visually group and compare selected segments of data.
• 200+ formatting settings to tailor the chart to different reporting and branding needs.

Figure: Drill Down Bubble PRO demonstrates multi-level drill down, panning, and zoom controls for exploring data from high-level patterns to detailed points.

These capabilities make Drill Down Bubble PRO especially useful for customer segmentation, sales analysis, portfolio analysis, performance monitoring, and other scenarios where users need to actively explore patterns rather than simply display them.

Commenting in view mode: introducing Zebra BI Tables+ & Charts+

It's month-end. A regional manager opens the report, spots a variance in their region, and knows exactly why it happened. But the report isn't theirs to edit, so the explanation ends up where it always does: a screenshot pasted into PowerPoint, emailed around, detached from the number that prompted it.

That's the collaboration gap. Your report shows what happened; the why and the what next leave the report and get lost.

Meet Zebra BI Tables+ and Charts+: two new visuals that keep everything existing Zebra BI Tables and Charts offer and add a built-in collaboration layer. Report viewers can add comments directly on a data point, in view mode, with no Power BI edit access. Each comment is anchored to the exact figure and stored in your own SharePoint file. Zebra BI stores nothing.

Figure: Zebra BI Tables+ and Charts+ enable report viewers to add contextual comments to individual data points, attach supporting files from SharePoint, and collaborate directly in view mode.

So instead of screenshotting insights into PowerPoint or maintaining a custom workaround, the whole team explains and decides right on the report. Instead of slides or emails, the next steps can be defined directly on the spot.

Stop losing good decisions to a missing context. Start making them on the report. Try Zebra BI Tables+ and Charts+ free for 14 days on AppSource.

Network Graph by Powerviz

The Powerviz Network Graph is a visual representation of Nodes and the relationships or connections between them. Its primary purpose is to illustrate how different entities are related or interact with one another.

Key Features

• Line Options: Customise node links with different styles and colors.
• Node Shape: Choose from various shapes, icons, or upload your own.
• Labels: Display categories, values, and percentage labels.
• Data Colors: Access 30+ color palettes, including color-blind-safe options.
• Fill Patterns: Highlight nodes with patterns or upload your own.
• Conditional Formatting: Highlight outliers based on measures or category rules.
• Clusters: Group related nodes to reveal patterns.
• Relationships: Analyze node connections at a glance.
• Animation: Add pop-ups and highlight effects.
• Import/Export: Save and reuse themes easily.
• Toolbars: Gain advanced control over the graph.

Other features included are Sorting, Ranking, Lasso / Reverse Lasso, Grid View and more.

Business Use Cases

Customer Relationship Management (CRM), Supply Chain Optimization, Social Network Analysis (SNA).

Figure: Customize every connection, cluster, and node to build a Network Graph that truly fits your data.Figure: Go beyond connections with powerful colors, patterns, labels, formatting, and interactive controls.Hierarchical Bar Chart

New, exciting changes were made to Hierarchical Bar Chart.

• Bubble Matrix for stacks - The visual now provides the option to show stacks as bubbles below the bars.

Figure: Hierarchical Bar Chart transforms stacked period values into bar-and-bubble matrix layouts for clearer comparison of category totals and component contributions.

• Display Stacks as Clustered Columns - From the format pane settings, you can choose to display stacks as clustered column chart.

Figure: Hierarchical Bar Chart displayed as clustered columns, with period targets and variance indicators configured from the Format visual pane.

• Filter using legends - You can click on a stack legend to filter out other stacks. On mouse hover relevant stack gets highlighted.
• Stack ribbons and %age of total - You can now choose to connect stacks with ribbon and show percentage of total for each stack.