Google Cloud's console model list recently gained a new entry: gemini-3.2-flash-lite-live-preview. This is the latest sighting of the Gemini 3.2 Flash series after earlier hints in iOS build packages and AI Studio earlier this month.
https://twitter.com/AiBattle_/status/2055760108693397644
The lite and live suffixes suggest Google is carving out a specialized version for ultra-low-latency real-time interactions. According to Abacus.AI CEO Bindu Reddy, the Gemini 3.2 Flash model matches about 92% of GPT-5.5's coding and reasoning capabilities — but thanks to distillation and sparsification techniques, its inference cost is just one-twentieth of GPT-5.5's, with most queries completing in under 200 milliseconds.
With the cloud API already jumping the gun, industry watchers expect this aggressively priced lightweight model to officially launch at Google I/O on May 20.