Google DeepMind releases Gemini 2.5 with native Figma plugin API

Tech · 5 min read

Google DeepMind's Gemini 2.5 update tightens integration between LLM capabilities and UI design tooling. The model now exposes a Figma-native API to read component hierarchies, constraints, and auto-layout configurations, enabling more semantically aware generation and refactoring inside design files.

Gemini 2.5 introduces a 'design intent' token that can be embedded into design documents to signal business rules like responsive priorities, localization targets, and component reuse. This allows the model to propose changes while honoring system constraints and developer handoffs.

DeepMind showcased demos where teams can ask Gemini to convert a desktop layout into a prioritized mobile pattern with preserved component variants and automated tokens for spacing and typography. The company also rolled out SDKs for Adobe XD and Framer, signaling cross-tool ambitions.