Google DeepMind launched Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking, two audio-to-audio models for voice agents that handle text, images, audio and video, and support tool calls during conversations. Extended Thinking focuses on long, multi-step tasks, continuing to speak while performing background reasoning or waiting on asynchronous tools, which forces developers to track interaction status and design non-blocking tool workflows. Both models, built on Gemini 3 Pro with a January 2025 knowledge cutoff, are widely distributed across Google products, watermarked with SynthID, and aimed at making real-time voice agents more capable without feeling unresponsive.
This update represents a notable development in the Board sector. Organizations and founders tracking this space should evaluate potential strategic and technical implications on their operations.