Google launches Gemini 3. 8 Live AI models for real-time voice interaction

Gemini 3. 8 Live voice AI agents multimodal AI real-time reasoning Gemini Notebook

Google has announced the release of its most advanced speech-to-speech AI models, Gemini 3. 8 Live and Gemini 3. 8 Live Extended Thinking. These models are designed to facilitate natural, low-latency voice interactions by allowing the AI to reason and speak simultaneously. While the standard Gemini 3. 8 Live focuses on scalability and cost-efficiency, the Extended Thinking variant is optimized for complex, multi-step reasoning tasks.

Key features include the ability to process visual information in near real-time, automatic detection and switching between 97 supported languages, and the capacity to execute API calls and tool functions in the background without interrupting the conversation. To combat misinformation, Google is implementing SynthID, an invisible digital watermark, on all audio generated by these models.

In addition to the core models, Google is updating Gemini Notebook (formerly NotebookLM) with new learning tools. These updates include real-time voice interaction in nearly 100 languages, an integrated audio recorder for capturing lectures, and the ability to generate interactive learning overviews, such as quizzes and flashcards, based on user-provided documents.

Entities

Artificial Analysis · Gemini · Google · Google DeepMind · OpenAI

Claims

What the coverage asserts, and how many sources carry each claim.

[● 6 SOURCES] Gemini 3. 8 Live Extended Thinking ranked first in the Speech to Speech Quality Index with a score of 82.6. tw.stock.yahoo.com · www.itmedia.co.jp · www.watch.impress.co.jp · www.techpluto.com · www.newsbytesapp.com · +1 more
[● 3 SOURCES] Gemini Notebook has added real-time voice interaction and audio recording features in nearly 100 languages. www.frandroid.com · japan.cnet.com · news.mynavi.jp
[● 4 SOURCES] The new Gemini models support automatic language detection and switching across 97 languages. tw.stock.yahoo.com · www.itmedia.co.jp · www.watch.impress.co.jp · www.newsbytesapp.com
[● 11 SOURCES] Google released Gemini 3. 8 Live and Gemini 3. 8 Live Extended Thinking models. tw.stock.yahoo.com · www.itmedia.co.jp · www.watch.impress.co.jp · www.techpluto.com · sotutto.it · +5 more
[● 3 SOURCES] Google is using SynthID to add invisible watermarks to all AI-generated audio to prevent misinformation. tw.stock.yahoo.com · www.watch.impress.co.jp · cryptonomist.ch
[● 3 SOURCES] The Extended Thinking model can perform reasoning and speech simultaneously to avoid conversation interruptions. www.itmedia.co.jp · sotutto.it