Google DeepMind demonstrated new real-time voice and vision features for the Gemini API at the AI Engineer World's Fair. Developers can now build low-latency multimodal agents using native audio processing across 90 languages, stateful tool calling, and live translation.
FireBrief
FireBrief