News
Google introduces Gemini with live video analysis and switching between 97 languages
The new Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking voice models work with video streams, external tools, and multilingual conversations in real time.
Google has introduced the Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking voice models for real-time interaction. VECTOR reports this. The models can analyze images and video streams during a conversation, as well as switch between 97 languages without requiring the dialogue to end.
Gemini 3.8 Live supports the use of external tools, expanding the capabilities of voice-based scenarios beyond ordinary messaging. Both models can also generate audio with a SynthID watermark—a label that makes it possible to identify AI-generated content.
Access to the new capabilities is available through the Gemini API and Google AI Studio. In addition, they are being integrated into Gemini Enterprise, Search Live, Gemini Live, and Workspace.
For Ukrainian musicians, producers, and teams, this could be useful for multilingual communication, material localization, working with video, and automating operational tasks. API access also makes it possible to test Ukrainian-language voice scenarios and create custom solutions, although the announcement does not describe any specific music integrations. When working with generated audio, the presence of the SynthID label should be taken into account.