AI Conversations Without Typing

Gemini 3.8 Live is designed around a more natural conversational experience. Instead of typing individual questions and waiting for a text response, users can communicate with Gemini using their voice and continue the conversation in real time.

Google says the new models are designed to handle interruptions and ongoing dialogue, making interactions feel closer to a continuous conversation rather than a sequence of individual prompts.

Gemini Can See What You See

One of the key capabilities of Gemini 3.8 Live is real-time visual understanding. Users can provide visual context through their device camera while talking with the AI.

This makes it possible to ask questions about objects, environments or situations visible through the camera while maintaining the conversation through voice.

For example, a user could point a camera at an object and ask Gemini to identify it, explain how something works or provide guidance based on what is currently visible.

Support for Multiple Languages

Gemini 3.8 Live can automatically detect and switch between supported languages during a conversation. Google says the system supports 97 languages and can transition between languages without requiring the user to manually change the conversation settings.

This capability is particularly relevant for multilingual conversations where users naturally move between different languages during the same interaction.

Tasks Can Continue in the Background

Another important feature is the ability to execute tools and API calls while the conversation continues.

Instead of forcing users to wait silently while Gemini completes a task, the system can acknowledge the request, continue communicating and perform certain operations in the background.

This approach is intended to make voice-based interactions more useful for complex workflows that require multiple steps.

Gemini 3.8 Live Extended Thinking

Alongside Gemini 3.8 Live, Google introduced Gemini 3.8 Live Extended Thinking. This version is designed for tasks that require deeper reasoning and multiple steps.

Extended Thinking can continue reasoning while maintaining the conversation, providing verbal progress cues while longer operations are being completed.

The result is intended to combine more advanced reasoning with the responsiveness of a live voice conversation.

Available Across Google's AI Ecosystem

Google is making the new Live models available across several of its products and services.

• Google Search through Search Live
• The Gemini mobile experience
• Google Workspace
• Gmail and Google Keep for supported subscribers
• The Gemini API
• Google AI Studio

Developers can also use the Gemini Live API to build voice-driven applications and interfaces around the new models.

From Search Box to Conversation

The introduction of Gemini 3.8 Live reflects a broader shift in how people interact with AI. Instead of treating an AI assistant as a text box that requires carefully written prompts, voice interfaces allow users to communicate more naturally and refine their requests as the conversation develops.

The addition of real-time visual understanding makes the interaction even more contextual, allowing the AI to work with information coming directly from the user's surroundings.

What Gemini 3.8 Live Could Mean for AI Assistants

Real-time voice interaction, visual context and background task execution could make AI assistants useful in situations where typing is inconvenient or where information needs to be understood as it happens.

From troubleshooting and learning to everyday questions and more complex workflows, Gemini 3.8 Live moves the interaction model closer to an ongoing conversation rather than a traditional search session.

Google is positioning the technology as a foundation for more capable voice agents, while developers can access the underlying models through Google's AI development platforms.

Availability

Google began rolling out Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking in September 2026. Availability varies depending on the product, subscription and use case.

Developers can access the models through the Gemini API and Google AI Studio, while consumers can experience the technology through Search Live and supported Gemini products.