Use itGemini 3.8 Live and 3.8 Live Extended Thinking, in the API today
Both are available now through the Gemini API and Google AI Studio at $0.005 per minute of audio input and $0.018 per minute of audio output.3 For a voice agent the useful part is asynchronous function calling — tool calls run in the background while audio keeps streaming — alongside visual grounding, mid-conversation switching across 97 languages, and alphanumeric parsing for confirmation codes.34
MATTERS TO · If you have been stitching a cascaded stack out of a recogniser, a model and a voice, this collapses three hops into one, with Live API integrations through LiveKit, Pipecat and Vercel.[^3] Enterprise access is still private preview.
WHAT THE VERDICT MEANS
Available, works, worth your time today.
Sources for this entry01 TypeSafe AI — Introducing System One Models & Jev, 15 Sep 2026 (2026-09-16) · 02 GitHub Changelog — SHA-1 in HTTPS on GitHub sunset, 15 Sep 2026 (2026-09-16) · 03 Google — Build real-time voice applications with Gemini 3.8 Live, 15 Sep 2026 (2026-09-16) · 04 Google — Gemini 3.8 Live & 3.8 Live Extended Thinking, 15 Sep 2026 (2026-09-16) · 05 Anthropic Status — error spikes, Claude Mythos 5.1 and Fable 5.1, 15 Sep 2026 (2026-09-16) · 06 arXiv 2609.17394 — Coding Agents Have Converged, 15 Sep 2026 (2026-09-16) · 07 Strix — admin access to Baseten GitHub repos, 1 Sep 2026 (2026-09-16) · 08 9to5Google — Gemini 3.8 Live announced, 15 Sep 2026 (2026-09-16) · 09 OpenAI Status — elevated errors, gpt-image-2.5-flare, 15 Sep 2026 (2026-09-16) · 10 Hacker News — Introducing System One Models and Jev, 1,085 points (2026-09-16)