Google Launches Gemini 3.8 Live, Extended Thinking — SkimNews
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Google launched Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking, two models built for near real-time voice reasoning to enable production-ready voice agents across the Gemini app, Google Workspace, and Search.
- Gemini 3.8 Live Extended Thinking took the #1 spot on Artificial Analysis' Speech to Speech Quality Index with an 82.6 score, led agentic task completion with 68.6% on τ-Voice and 35.1% on Sierra's τ-Voice-banking benchmark, and scored 97.7% on Big Bench Audio.
- Gemini 3.8 Live placed second in Speech Agent Arena, and on ServiceNow's EVA-Bench both models pushed the Pareto Frontier for complex workflows by balancing accuracy with conversational quality.
- Both models process visual inputs in near real-time, auto-detect and transition between 97 supported languages mid-conversation, and run tools and API calls in the background while the conversation continues; Extended Thinking can reason and speak simultaneously with verbal cues like "Let me check that…"
- All AI-generated audio from the models is watermarked with SynthID to keep AI content detectable and help prevent misinformation.
- Developer platforms Agora, Fishjam, LangChain, LiveKit, Pipecat, Vercel, and Vision Agents are integrating via the Gemini Live API, while partners Salesforce, Genspark, and Lumeris cited low latency, fluidity, and tool-calling.
- 3.8 Live is rolling out today in the Gemini API, Google AI Studio, Gemini Enterprise (private preview), and Search Live; Extended Thinking also launches today in Gemini API, AI Studio, Gemini Enterprise and Customer Experience (private preview), Google Workspace business, Gemini Live, Docs, Gmail, and Keep for Google AI Pro/Ultra subscribers.
Why it matters: Developers building voice agents now have a Google-native option claiming top benchmark scores (#1 Speech to Speech Quality at 82.6, 68.6% on τ-Voice, 97.7% on Big Bench Audio) shipping today across the Gemini API, AI Studio, Gemini Enterprise, and Search Live — with SynthID watermarking on all audio for transparency. The 97-language mid-conversation auto-detection and background tool-calling directly target the friction points enterprise voice-agent deployments hit in production.
Ask SkimNews
