Google DeepMind has announced that developers can begin building with two models: Nano Banana 2 Lite and Gemini Omni Flash. The announcement, posted on the DeepMind blog, is framed as an invitation to start working with the models rather than as a detailed technical specification.
Because this analysis draws on a single source, there are no independent details about performance, benchmarks, or pricing. The post's title emphasizes "start building," suggesting the focus is on early access and developer adoption rather than on head-to-head model comparisons.