Google Expands Gemini 3.8 Family With Live and Live Extended Thinking

Google is expanding its Gemini 3.8 artificial intelligence lineup with two new voice-focused models, Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking. The additions are designed to support more natural real-time conversations while addressing different performance and computing needs.

The expansion comes roughly two weeks after Google introduced Gemini 3.8 Flash, which itself followed Gemini 3.7 Flash by only about three weeks. Rather than introducing another direct successor, Google is broadening the Gemini 3.8 family with models specifically designed for interactive voice experiences.

Gemini 3.8 Live Targets Scale and Cost Efficiency

Gemini 3.8 Live is designed for production environments where scalability and cost efficiency are priorities. Google says the model is intended to make voice assistants more responsive and collaborative while handling multiple types of information during a conversation.

One of its key capabilities is near-real-time processing of visual inputs. This could allow applications powered by Gemini to analyze what a camera or other visual source is capturing while continuing a spoken interaction with the user.

The model can also recognize transitions between 97 supported languages during a conversation. That capability is intended to make multilingual exchanges more seamless, particularly when users switch languages without explicitly telling the assistant to do so.

Gemini 3.8 Live can also manage tools and API calls in the background. According to Google, this allows the AI to acknowledge a user’s request and continue participating in the conversation while completing other operations behind the scenes.

Live Extended Thinking Focuses on Complex Tasks

Gemini 3.8 Live Extended Thinking is aimed at more demanding workloads that require additional reasoning. Unlike conventional voice systems that may separate reasoning from speech generation, Google says the model can reason and speak simultaneously.

That approach is designed to reduce interruptions during complex conversations and allow the assistant to continue communicating while working through a problem.

Google says Gemini 3.8 Live Extended Thinking reached the top position on Artificial Analysis’ Speech to Speech Quality Index. The company also points to strong results in agentic task completion and τ-Voice-banking.

Banking Benchmark Tests Real-Time AI Agents

The τ-Voice-banking benchmark evaluates audio-native conversational AI agents on complex banking and financial technology customer-service tasks.

Such tests are intended to measure more than basic speech recognition or response quality. They examine whether an AI system can understand spoken instructions, reason through multi-step requests and successfully complete tasks in scenarios resembling real-world customer interactions.

Google’s emphasis on these capabilities reflects a broader effort across the AI industry to develop assistants that can move beyond simple question-and-answer exchanges and perform more complicated actions while maintaining natural conversations.

SynthID Watermarks AI-Generated Audio

Audio produced by both Gemini 3.8 Live models will include Google’s SynthID watermarking technology.

SynthID embeds signals into AI-generated content that are designed to remain imperceptible during normal use while helping identify material created by artificial intelligence. Google has expanded the technology across several types of generative media as companies seek ways to distinguish synthetic content from human-created material.

The inclusion of SynthID in Gemini 3.8 Live and Live Extended Thinking means audio generated through the new systems will carry Google’s digital watermark.

Gemini 3.8 Live Models Roll Out Across Google Services

Both models are beginning their rollout across Google’s consumer products.

Gemini 3.8 Live will be available through Search Live, bringing its real-time conversational capabilities into Google’s search experience.

Gemini 3.8 Live Extended Thinking, meanwhile, will be offered through Gemini Live for users who need more advanced reasoning during voice interactions.

Google is also expanding access for subscribers. Users with AI Pro and Ultra subscriptions will be able to use Live Extended Thinking in Google Docs, Gmail and Keep, extending the model’s conversational and reasoning capabilities into Google’s productivity ecosystem.

With Gemini 3.8 Live emphasizing scalable, cost-efficient voice interactions and Live Extended Thinking targeting more complicated reasoning tasks, Google is broadening the Gemini platform beyond traditional text-based AI. The rollout also signals the company’s continued push toward assistants capable of handling voice, visual information, tools and multi-step tasks within a single real-time conversation.

Ernest Hemingway

Ernest Hemingway is a contributor at Thecherawchronicle.com, covering a wide range of topics including news, politics, business, technology, sports, entertainment, and lifestyle. He focuses on delivering clear, balanced reporting and useful information that helps readers stay informed about current events and issues that matter to their communities. His work emphasizes accuracy, relevance, and accessibility, bringing readers timely stories and practical insights in a straightforward and engaging way.