Google Unveils Gemini 3.8 Live Models with Simultaneous Thinking and Speech
Google has launched Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking, introducing real-time simultaneous processing and speech capabilities that eliminate assistant delays.
Google has announced a major breakthrough in real-time artificial intelligence with the launch of two new models: Gemini 3.8 Live and the flagship Gemini 3.8 Live Extended Thinking, eliminating the awkward silence previously experienced during complex voice assistant queries.
Thinking and Speaking Simultaneously
The primary engineering breakthrough of the Extended Thinking model is its ability to process information and converse simultaneously. Instead of halting the conversation, the model uses natural verbal cues such as "Let me check that for you," continuing to speak while running complex background tasks. Furthermore, the system provides real-time verbal updates on task progress, allowing users to redirect or refine instructions on the fly without waiting.
These capabilities integrate deeply into Google Workspace services. Users can leverage Gmail Live for intelligent voice searches, Docs Live to draft documents pulling data from web files and Drive, and Keep Live for voice-activated notes and shopping lists. The system can coordinate restaurant reservations, troubleshoot technical searches, and generate code and applications seamlessly.
Global Language Support and Performance
Alongside the extended reasoning model, Google introduced the base Gemini 3.8 Live model, optimized for speed and cost-efficiency. It processes real-time camera visual inputs, analyzes images, and plays chess through continuous dialogue. A standout feature for global audiences is the automatic detection and switching between 97 languages mid-conversation. In global benchmarks, the Extended Thinking model secured first place in Artificial Analysis's speech quality index with a record score of 82.6.
For security, Google embedded a transparent SynthID digital watermark within the generated audio signals to ensure transparency and prevent deepfakes. The models are rolling out starting today across Google AI Studio, Gemini Enterprise, and Search Live, with consumer access expanding via the Gemini app and Google AI Pro and Ultra subscriptions.
