Power BI TMDL View on the Web 2026: Edit Semantic Models as Code in Your Browser

Power BI’s semantic model – the tables, relationships and DAX behind your reports – has traditionally been edited in a visual designer inside Power BI Desktop. In 2026 Microsoft brought TMDL View to the web, giving you a proper code editor in the browser where you can see and edit a model as text. For anyone who has wanted version control, repeatability and AI assistance for model changes, this is a meaningful shift in how professional Power BI work gets done.

What TMDL actually is

TMDL stands for Tabular Model Definition Language – a human-readable, text-based way to describe a semantic model: its tables, columns, measures, relationships and properties. Instead of clicking through dialog boxes, you read and write the model as code. That has two big advantages: changes become reviewable and diff-able like any other code, and large or repetitive edits can be made in bulk by editing text rather than one property at a time.

Editing models in the browser

TMDL View on the Web brings a rich code editor directly into the Power BI service. You can open a semantic model, view its definition as TMDL scripts, edit them, then preview and apply the changes – all without installing or opening Desktop. For teams on managed laptops or working across machines, removing the Desktop dependency for model tweaks is genuinely useful. It also makes the model’s structure visible and auditable in a way the visual designer never quite managed.

AI assistance with Copilot and GitHub Copilot

Because the model is now text, AI tooling can help author it. You can use AI assistants – including GitHub Copilot – to generate or modify TMDL, then paste it back into the editor to preview and apply. Separately, Copilot in web modeling lets you analyse and improve a model in natural language: renaming tables and columns, creating relationships and generating DAX measures by describing what you want. The combination means a lot of tedious modelling work can be drafted by AI and refined by you.

Why code-first modelling matters

Treating a model as code unlocks the disciplines software teams take for granted: source control, code review, and consistent deployment across development and production. A measure defined once can be copied, templated and reused rather than rebuilt. Mistakes are easier to spot in a diff than in a dialog. As semantic models grow to hundreds of measures, this is the difference between a maintainable asset and an unmanageable one.

Getting started safely

Because TMDL View on the Web is in preview, treat it the way you would any preview feature: experiment on a copy of a model, not your production report’s live dataset. Learn to read TMDL before you write much of it – generate a script from an existing model and study how tables and measures are expressed. Keep changes small, preview before applying, and validate your reports still render correctly afterwards. Used carefully, it is the most powerful way yet to build and maintain Power BI models at scale.

Key takeaways

• TMDL (Tabular Model Definition Language) describes a Power BI semantic model as readable text rather than dialog boxes.
• TMDL View on the Web puts a full code editor in the Power BI service – edit models without opening Desktop.
• AI tools including GitHub Copilot can generate or modify TMDL, and Copilot in web modeling edits models in natural language.
• Code-first modelling enables source control, code review and consistent, repeatable deployments.
• It is in preview – experiment on copies, preview before applying, and validate reports afterwards.

Featured video: Fabric Insider Ep 5: What's Hot and New for Power BI Semantic Models in 2026 (Christian Wade)
Related sources: TMDL View on the Web (Preview) – Microsoft Fabric Community · Use Copilot with Semantic Models in Power BI – Microsoft Learn · Power BI June 2026 Feature Summary – Microsoft Fabric Community