Google has launched two new Gemini voice AI models designed to make conversations with artificial intelligence faster, smoother, and more natural. The new models, called Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking, are built for real-time voice interactions.
Gemini 3.8 Live is designed for fast conversations where users can speak naturally with AI and receive responses with very little delay. Google says the model can continue a conversation while also carrying out other tasks in the background. This could make voice-based AI assistants more useful for customer service, workplace tools, productivity apps, and other services.
The model can also use visual information during a conversation. This means an AI assistant could understand what a user is showing through a supported camera or visual input while also listening to spoken questions. It can then use both types of information to provide a more useful response.
Another major feature is multilingual support. Gemini 3.8 Live can work across 97 supported languages and can automatically switch between languages during the same conversation. This could make voice AI more practical for users and businesses operating across different countries.
For more difficult tasks, Google has introduced Gemini 3.8 Live Extended Thinking. This version is designed to handle problems that require deeper reasoning or several steps before reaching an answer.
Instead of becoming silent while working on a complicated request, the Extended Thinking model can continue speaking while reasoning in the background. It can give users short updates as it works through different parts of a task, creating a conversation that feels more continuous.
The new models also support background tool and API calls. For example, a voice assistant could search for information, interact with another service, or complete a digital task without completely stopping its conversation with the user.
Google is making the models available to developers through the Gemini API and Google AI Studio. This will allow developers to create real-time voice assistants and other speech-based applications using Gemini technology.
Gemini 3.8 Live is also being introduced across some Google products and services, including Search Live. Gemini 3.8 Live Extended Thinking is being introduced through Gemini Live and selected Google Workspace experiences, with availability depending on the product and subscription.
The launch shows how AI companies are moving beyond traditional text chatbots toward systems that can listen, speak, reason, see visual information, and perform tasks at the same time.
Real-time voice interaction is becoming an important area of AI development because it could make digital assistants easier to use without keyboards or screens. Google’s latest Gemini models are aimed at making these conversations feel faster and more natural while giving AI systems stronger reasoning and task-completion abilities.
