On September 15, 2026, Google introduced Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking—near real-time voice models for voice agents and fluid dialogue. The companion developer post prices Live API audio at $0.005/min input and $0.018/min output. Those figures are USD as published. CAD is on the invoice. This desk is not inventing a Canadian list price.
What Google listed for Gemini 3.8 Live
Google frames Gemini 3.8 Live as built for scale and cost efficiency, combining conversational intelligence with fluid dialogue and visual grounding. It automatically detects and transitions between 97 supported languages mid-conversation, and executes tools and API calls in the background while continuing the conversation. The developer post adds asynchronous function calling, visual context, alphanumeric precision, 97+ languages, and incremental content updates.
Gemini 3.8 Live Extended Thinking is framed for high-complexity, multi-step reasoning. Google says it reasons and speaks simultaneously, uses early verbal cues like “Let me check that…” to acknowledge prompts, and uses live progress narration for multi-step background tasks. Configurable thinking is noted on the developer post for complex multi-step work while narrating progress in the main conversation.
Google-reported benchmarks (vendor figures, not this desk’s independent run): Extended Thinking #1 on Artificial Analysis Speech to Speech Quality Index at 82.6; 68.6% on τ-Voice; 35.1% on Sierra’s τ-Voice-banking; 97.7% on Big Bench Audio. Live is described as second place in the Speech Agent Arena. Audio generated by Google’s AI products is watermarked with SynthID.
“For tasks that require deeper reasoning, 3.8 Live Extended Thinking reasons and speaks simultaneously. It delivers increased intelligence for complex workflows while maintaining an uninterrupted conversational flow — using early verbal cues like “Let me check that…” to acknowledge prompts naturally, and live progress narration to walk users through multi-step background tasks as they progress.”Google blog: Introducing Gemini 3.8 Live and 3.8 Live Extended Thinking, September 15, 2026 (Updated September 17, 2026)
What it costs
These are Google’s published USD Live API audio rates from the developer post. Do not quote them as CAD. CAD is on the invoice. The footnote states the per-minute figures are an estimate based on $3/1M tokens for input and $12/1M tokens for output.
| Line | Rate | Billing | Notes |
|---|---|---|---|
| Audio input | $0.005 / min | Live API | Estimate from $3/1M tokens in |
| Audio output | $0.018 / min | Live API | Estimate from $12/1M tokens out |
| Partners | — | — | Agora, Fishjam, LangChain, LiveKit, Pipecat, Vercel, Vision Agents |
| Compare (our desk) | $0.05 / min | GPT-Live-1 | OpenAI voice layer from our Live-1 briefing |
Rollout Google published
Google’s Sep 15 post (updated Sep 17) lists rollout as follows — exact surface bullets, not invented:
3.8 Live
- For developers: In the Gemini API and Google AI Studio
- For enterprises: In private preview in Gemini Enterprise and coming soon to Gemini Enterprise for Customer Experience
- For everyone: In Search Live
3.8 Live Extended Thinking
- For developers: In the Gemini API and Google AI Studio
- For enterprises: In private preview in Gemini Enterprise and coming soon to Gemini Enterprise for Customer Experience and Google Workspace business customers
- For everyone: In Gemini Live and for Google AI Pro and Ultra subscribers in Workspace in Docs, and all Google AI subscribers in Gmail and Keep
What Alberta operators should do
Alberta work here means AI chatbot lead generation and AI automation Alberta—phone-line receptionist, missed-call capture, and booking flows for shops that lose leads when nobody picks up. Gemini 3.8 Live’s Live API pricing ($0.005/min audio in, $0.018/min audio out) is the operator cost card for a voice front door that can keep chatting while tools run in the background. Extended Thinking is the path when the call needs multi-step reasoning without dropping the conversation.
For same-currency USD cost context only, our GPT-Live-1 briefing lists OpenAI’s voice layer at $0.05/min—about ten times Gemini’s published audio-input per-minute rate. That is not a feature bake-off; it is operator budget math. Foreign hosted API is not a private Alberta path. For regulated data—client health notes, tenders, credentials—prefer a private AI security stack with a human review gate before anything leaves the shop. If the workflow has to stay on the desk with clear ownership, that is AI consulting work, not a public paste into a third-country endpoint. Quote USD as published; expect CAD on the invoice.
The guardrail
Price is not permission. A $0.005-per-minute audio-input card does not put inference in Alberta, and it does not authorize shipping regulated client audio offshore without a written data boundary. For AI chatbot lead generation, use Gemini 3.8 Live when the job fits a real-time voice agent with background tools, use Extended Thinking when the call needs live reasoning narration, keep a human in the loop on bookings and care-adjacent intents, and keep sensitive workloads on a private agents path. SynthID watermarking is Google’s transparency mark on AI audio—not a substitute for your own consent and retention policy.