Google DeepMind has introduced Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking, positioning them as its most capable conversational AI models and adding background task handling to the company’s assistant strategy. The announcement could give Google a stronger answer to rivals pursuing more continuous, action-oriented AI.

From chatbot to persistent assistant

In a post on X, Google DeepMind said the new models can talk with users, reason through requests and handle tasks in the background without breaking the flow of a conversation. The company described both versions as its best conversational AI, but did not provide pricing, availability, performance benchmarks or a detailed account of which tasks they can complete.

That missing detail is significant. Background processing is potentially more important to the user experience than another incremental improvement in answer quality. If Gemini can continue working while a person keeps speaking, the assistant could become less dependent on rigid turn taking. Users might not need to wait in silence, restart a session or divide a complex request into short exchanges.

The commercial value will depend on execution. A system that quietly handles useful work could increase engagement and make Gemini more central to daily tasks. A system that starts unclear or unauthorized actions could create trust problems, particularly if users cannot see, pause or cancel what is happening.

Google’s competitive position

The launch fits a broader industry effort to turn AI assistants into persistent interfaces rather than simple question and answer tools. OpenAI, Anthropic and other competitors are also seeking to combine natural conversation with reasoning, tool use and longer running workflows. Google’s advantage is its position across search, mobile software, cloud services and consumer applications, giving it multiple routes for distributing a more capable assistant.

Gemini 3.8 Live Extended Thinking suggests Google is also trying to close the gap between conversational speed and analytical depth. Voice and multimodal interaction have often been treated as lightweight features, while difficult reasoning has required a slower, more deliberate mode. Combining the two could make the assistant more useful for planning, research and other complex work.

Still, the announcement establishes a direction rather than proving a market advantage. Google must show what “background” means in practice, how reliably the models distinguish conversation from action and when they require explicit approval. Until those limits are disclosed, Gemini 3.8 Live is best understood as a strategic signal: Google wants the assistant to remain present, productive and commercially embedded beyond the individual prompt.

#Google DeepMind#Google#Gemini 3.8 Live#Gemini 3.8 Live Extended Thinking#Gemini#OpenAI#Anthropic
About Rebeca Smith
Rebeca Smith is an AI and technology journalist specializing in the business of artificial intelligence. Her reporting focuses on the companies, investments, and competitive strategies driving the industry's rapid evolution. She closely follows Big Tech, AI startups, venture capital, semiconductor manufacturers, and enterprise software, explaining how commercial decisions shape the future of AI adoption. Rebeca's work combines financial insight with technological understanding, helping readers see beyond product launches to the economic forces transforming the industry.

This article was written with the assistance of an AI system and published automatically.