Get App
Download App Scanner
Scan to Download
Advertisement

Google Launches Gemini 3.8 Live, New Extended Thinking Voice AI Models

Google’s new Gemini 3.8 Live models expand real-time voice AI with asynchronous function calling, visual inputs and extended thinking capabilities.

Google Launches Gemini 3.8 Live, New Extended Thinking Voice AI Models
Google Launches Gemini 3.8 Live and Extended Thinking Voice Models
Image: AI generated

Google has launched two new voice-focused AI models, Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking, there aim is to make real-time conversations with AI more natural while allowing the models to reason and use tools in the background.

Announced on September 15, the models are designed for voice-based AI agents that can continue speaking with users while completing tasks, rather than pausing the conversation until a tool call is finished.

Gemini 3.8 Live supports asynchronous function calling, allowing it to run APIs and other tools in the background while continuing to stream audio. It can also process live visual inputs and handle alphanumeric information such as confirmation codes and claim numbers with greater accuracy.

Also Read | Layoffs Are Rising In Tech, But These Skills Are Helping People Unlock Better Pay

Google said the model supports more than 97 languages and can maintain accent consistency when switching between languages.

Alongside the new voice models, Google has introduced Gemini 3.5 Transcribe, a speech-to-text model built for accurate transcription. It supports more than 85 languages, including conversations involving multiple languages, and can handle specialised terms and alphanumeric information.

Its smart transcription mode can also remove filler words, account for self-corrections and automatically format transcripts, Google said.

Also Read: Google Launches Gemini App For Windows 10, 11

Extended Thinking Targets Complex Tasks

Gemini 3.8 Live Extended Thinking is designed for more complicated, multi-step tasks. It can reason in the background while continuing a conversation and can provide live progress updates as it works through a task.

Google said the model ranked first on Artificial Analysis' Speech-to-Speech Quality Index, with a score of 82.6, while also recording a 97.7% score on the Big Bench Audio benchmark.

For developers, the models are available through Google's Live API and are priced at $0.005 per minute for audio input and $0.018 per minute for audio output.

Alphabet Shares: Recent Movement

Despite the launch of the new models, shares of Alphabet Class A declined by 1.26% to $344.98 on Sept. 15.

During the past 52 weeks, the stock has traded between $235.84 and $408.61.

The stock was trading at a price-to-earnings multiple of 17.3 times, with a market cap of around $4.2 trillion as at the end of the previous trading session.

Also Read: OpenAI Says It's Working With Anthropic, Google On AI Safety

Essential Business Intelligence, Sharp Market Insights, Practical Personal Finance Advice, Daily Fuel, Gold and Silver Prices and Latest Stories — On NDTV Profit.

Newsletters

Update Email
to get newsletters straight to your inbox
⚠️ Add your Email ID to receive Newsletters
Note: You will be signed up automatically after adding email

News for You

Set as Trusted Source
on Google Search
Add NDTV Profit As Google Preferred Source
Listen to the latest songs, only on JioSaavn.com