The Gemini 3.2 real-time model is now available on Google Cloud, offering a cost-effective alternative to GPT-5.5. The model, named "gemini-3.2-flash-lite-live-preview," has been added to the Google Cloud Console, marking its public debut on an official platform. This follows earlier appearances in iOS app bundles and AI Studio. Gemini 3.2 Flash achieves 92% of GPT-5.5's coding and reasoning capabilities but at only one-twentieth of the inference cost, thanks to advanced distillation and sparsification techniques. Most query latencies are under 200 milliseconds, making it ideal for ultra-low-latency real-time interactions. Industry insiders expect its official launch at Google I/O on May 20.