Framer’s Agent (3D design capability) Adds Editable 3D Website Transforms.

Framer’s Agent can now create three-dimensional website effects without flattening the result into an image or locking designers out of the underlying work. The update gives the AI system control over perspective, depth, rotation and animation, while preserving each generated layer as an editable object on Framer’s canvas.

The company detailed the capability in an official product update, followed by a LinkedIn announcement. Framer said its Agent can coordinate properties across all three axes, including Rotate X, Y and Z, Perspective, Origin, Depth and Backface settings. That allows it to build effects such as dimensional logos, rotating cards and layered interface compositions from a prompt.

The more consequential part is editability. Generated designs use the same Frames and Effects available elsewhere on the canvas, so users can select individual layers, change their properties and continue refining the composition manually. This avoids a familiar limitation of generative design tools, where an impressive initial output can become difficult to adjust once a client requests a smaller rotation, different timing or revised copy.

Framer’s example is a dimensional animation of its own logo. According to the company, the composition consists entirely of editable Frames and Effects rather than an imported render. It was generated using Opus 5 Fast with Light reasoning.

The Agent’s model picker recently gained controls for reasoning and Fast mode. Light reasoning is intended for quick changes, while higher settings give a model more room to plan before building. Fast mode can reduce generation time on supported models, although Framer notes that it consumes more tokens per task. For the 3D capability, the company said Opus 5 performed particularly well.

A practical workflow could look like this:

• Ask the Agent to create a card stack with perspective and staggered depth.
• Have it add coordinated rotation and entrance animations.
• Select the resulting Frames on the canvas.
• Fine-tune origins, angles, timing or content without regenerating the entire design.

This is closer to assisted construction than image generation. The Agent is manipulating Framer’s native design system, which should make its output more useful for production sites and iterative client work.

3D web effects traditionally require a mix of transform knowledge, animation tooling and careful visual tuning. Framer’s update does not remove the need to test usability, responsiveness or performance, but it can shorten the path from an idea to a working composition. Designers can use natural-language instructions to establish the geometry, then take over for detailed adjustments.

The release also fits Framer’s broader push toward richer visual effects inside its website builder. It follows Framer Video Shaders bringing real-time effects to web video and the addition of Fable 5.1 for shaders, effects and interactions. Together, these features position the canvas as a place for producing more complex motion and depth without exporting work to a separate graphics application.

There are limits to the announcement. Framer has not described the feature as a full 3D modelling system, and the listed controls concern transforms applied to layers and frames. Users should therefore expect dimensional web compositions, not generated meshes or textured 3D assets. The company also did not disclose feature-specific pricing or plan requirements in the supplied announcement.

What is Framer’s Agent (3D design capability)?

It is an update that lets Framer’s AI Agent create designs using 3D transform properties such as perspective, depth, origin, backface settings and rotation across the X, Y and Z axes. The generated layers remain selectable and editable on Framer’s canvas.

Does Framer’s Agent generate editable 3D designs?

Yes. Framer says the Agent builds with native Frames and Effects, allowing users to modify individual layers and properties after generation. The feature is focused on editable 3D web effects rather than conventional 3D models or meshes.

Which model does Framer recommend for 3D transforms?

Framer said Opus 5 was especially effective for the capability. Its published logo example used Opus 5 Fast with Light reasoning, though the announcement does not state that this configuration is mandatory.