AI BriefingAI Gateway Begins Supporting Gemini 3.8 Live and Extended Thinking Models
Key point
AI Gateway has added support for Gemini 3.8 Live and Extended Thinking models, enabling real-time voice interactions.
Details
AI Gateway has begun supporting Google's Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking models. Both models enable real-time voice interactions in voice assistants, conversational experiences, and audio response applications.
Gemini 3.8 Live supports real-time audio, visual grounding, and automatic switching between 97 languages, and can perform tool calls in the background during conversations. Gemini 3.8 Live Extended Thinking performs multi-step reasoning in parallel with voice processing, so conversations are not interrupted even when acknowledging requests or explaining progress.
Developers can use both models via the AI SDK's real-time API. After installing the Gateway provider and WebSocket client, obtain a short-lived token, open a WebSocket connection, and serialize and parse real-time events through the model adapter. AI Gateway provides a unified API for model invocation, usage and cost tracking, and retry and failover configuration.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.
Related
Google Includes Premium Colab Benefits in AI Subscription Tiers