Gemini 3 Pro lands in Figma Make as an experimental model

Figma has made Gemini 3 Pro available as an experimental model in Figma Make. The team evaluated the new model across a range of design exercises before shipping it, with tests highlighting consistent performance in layout exploration, style variation, and interactive pattern generation. A lighter sibling, Gemini 3 Flash, is also available for rapid ideation and quick refinements.

Co-founder and CEO Dylan Field demos some of our favorite early experiments with Gemini 3 Pro.

From static design to code-backed prototype

One of the key tests focused on bridging the gap between a crafted design and working code. The goal was to see whether Gemini 3 Pro could convert a new design into an accurate, code-backed prototype without losing the feel of the original work.

As a trial, designers built a Thanksgiving gratitude board: a set of golden leaves on a deep maroon canvas in Figma Design. They asked Figma Make to generate SVGs and animate them with on-screen physics. The initial prompt produced a calm, organic flow of leaves across the screen. Adding a few more instructions and a Supabase connection turned the concept into an interactive experience where guests could submit gratitude notes, with each note becoming a leaf that revealed its message on hover.

A Thanksgiving gratitude board comes to life—golden leaves drift with realistic physics, revealing guest notes on hover.

Shifting aesthetics without losing fidelity

To test stylistic range, Figma used a New Year's Eve RSVP page as a playground. The feature set was simple, but the visual treatments were deliberately extreme.

One prompt asked for a Y2K treatment: retro-futuristic, moody dark chrome energy with a functioning RSVP form. Gemini 3 Pro delivered on both the visual tone and the interaction model. A second prompt veered in the opposite direction, requesting a "concrete poetry" approach—severe typography, brutalist tension, minimal ornamentation.

The model matched each shift without sacrificing fidelity. Each style remained distinct and coherent while the working form stayed intact. Motion and interactive details were generated consistently, matching the aesthetic of each variant.

For the concrete poetry variation, the team’s New Year’s Eve RSVP becomes text-as-structure—stark type, minimal ornament—while keeping the core interaction intact.

Working inside design systems

Beyond open-ended exploration, Figma tested Gemini 3 Pro in more constrained, production-like environments. The test used the design system foundations available in Figma Make through Make kits and npm imports.

Starting from a FigJam template built on the UI3 library, they asked the model to add a feature that lets users switch canvas backgrounds between different styles. Gemini 3 Pro generated 12 distinct styles, implemented them with UI3 components, and wired up proper interactivity with minimal instruction. It also improvised extra details, such as animating between textures and adding sticky note layering and scaling effects.

Gemini 3 Pro helped the team prototype a canvas background switcher concept in FigJam templates—12 styles, smooth transitions, all improvised.

A wider canvas for exploration

The consistent theme across these tests was that AI extends the designer's reach. As models like Gemini 3 Pro gain design fluency, the space for creative exploration widens. Speed no longer stands in opposition to creativity; instead, the designer's role in steering possibilities becomes more central.

To try Gemini 3 Pro or Gemini 3 Flash in Figma Make, open your settings in Figma, navigate to the experimental models section, and toggle to your preferred model.