Gemini 3.1 Flash-Lite now builds websites in real time

Google DeepMind's latest model, Gemini 3.1 Flash-Lite, can now generate complete websites almost instantly. It's no longer just text streaming out quickly — full user interfaces materialize directly in the browser as you watch.

Google recently published a pseudo-browser demo in AI Studio where users can type a text description of a webpage and watch it being built live, line by line, in real time. The output isn't always consistent and content can drift into nonsense without tight guardrails, but the potential is clear: rapid UI mockups, prototyping, and visual ideation can now happen in seconds.

The performance numbers are compelling. According to Google, the model is 2.5x faster than Gemini 2.5 Flash on time-to-first-token and pushes over 360 tokens per second. The tradeoff is a price that has more than tripled, from $0.40 to $1.50 per million output tokens. Still cheap in absolute terms, but a notable jump.

The model has been available in Google AI Studio and Vertex AI since early March, and according to Artificial Analysis, it outperforms larger models like Claude Opus 4.6 on some multimodal tasks.

For CIOs and development teams, this is worth watching. The speed alone opens up use cases that weren't practical before — real-time UI generation in self-service portals, rapid prototyping during product meetings, and automated landing page generation based on campaign parameters.