Microsoft launches Copilot 3D to turn photos into 3D models

Microsoft has unveiled Copilot 3D, an artificial intelligence-powered tool that converts regular 2D images into 3D models within seconds. The launch, announced a day after the company rolled out GPT-5-powered Smart Mode in Copilot, reflects Microsoft’s accelerating push to integrate AI into creative and design workflows.

The new feature is part of Copilot Labs and is aimed at making 3D creation “fast, accessible and intuitive,” according to Microsoft’s product description. It is currently available for free to a subset of users, with sign-in via a Microsoft or Google account required.

In its official documentation, Microsoft notes that Copilot 3D is designed for rapid prototyping, concept testing, and educational use — areas where traditional 3D modelling tools can be too complex or time-consuming.

The tool works by allowing users to upload PNG or JPG images under 10MB. Once an image is uploaded, pressing the “Create” button triggers the AI to generate a 3D version within a few seconds to a minute. The resulting models can be exported in GLB format, a widely supported 3D file type compatible with most 3D viewers, tools, and game engines.

According to The Verge, early tests show that Copilot 3D performs best with objects such as furniture or everyday items like umbrellas and bananas, though it may struggle with animals or more complex shapes.

How to access Copilot 3D

To try the feature, Microsoft advises using a desktop browser. Users can visit Copilot.com, click the sidebar button on the top left, navigate to “Labs,” and then select “Try now” under Copilot 3D.

Once models are generated, they are stored on the “My Creations” page for 28 days and can be exported to Augmented Reality (AR) applications. This temporary storage approach allows users to download and manage their files while controlling server-side retention.

Limitations and best practices

Microsoft recommends uploading images with a clear subject-background separation for best results. While current file support is limited to PNG and JPG formats, the company may expand compatibility and file size limits in future updates.

Users are urged to upload only images they own the rights to and to avoid submitting pictures of people, in line with Microsoft’s terms of service. The company warns that accounts may be suspended for violations, including uploading illegal content, which will be automatically blocked.

Microsoft also stated that these user-generated 3D models will not be used to train its AI systems, addressing a common privacy concern with generative AI tools.

Part of a broader AI push

The launch of Copilot 3D comes just one day after Microsoft integrated GPT-5-powered Smart Mode into Copilot, enabling more context-aware and accurate AI interactions. The back-to-back releases highlight Microsoft’s ambition to position Copilot as a multi-functional platform for productivity, creativity, and design.

Analysts note that the ability to quickly generate 3D assets could be particularly appealing to game developers, product designers, and educators — sectors where 3D content is in high demand but often resource-intensive to produce.

By lowering the technical barriers, Microsoft appears to be targeting not just professional creators but also casual users and learners who want to explore 3D design without mastering complex software such as Blender or Autodesk Maya.