Google and OpenAI released two sharply different approaches to solving voice agents' latency problem within just five days of each other, according to a report published by The New Stack. Google launched Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking through the Gemini API and Google AI Studio on Tuesday, five days after OpenAI released GPT-Live-1. Both systems allow a voice agent to continue speaking while processing background tasks, but their architectural designs diverge substantially.
Gemini 3.8 Live Extended Thinking keeps speech, reasoning, and tool execution inside a single stateful session, even while external API calls are still running. When a function is set to NON_BLOCKING, Gemini can continue speaking while it waits for the tool to respond, asking follow-up questions or providing updates. Developers can adjust reasoning effort to low, medium, or high per request. Standard Gemini 3.8 Live skips the extended reasoning step to reduce latency and token cost. GPT-Live-1 takes a different path, with the voice model handling the full-duplex conversation while a backend model such as GPT-6 Astra, a lighter model like Luna, or even a third-party option handles reasoning and tool execution independently. OpenAI puts turn-taking latency at around 800 milliseconds. The tradeoff is that developers must coordinate the two layers themselves, passing context between the voice model and backend reasoner through sideband channels.
The report notes that Google tells The New Stack that Extended Thinking handles "background noise, heavy accents, and unexpected interruptions better than competing models." Gemini 3.8 Live Extended Thinking scored 82.6 on Artificial Analysis' Speech-to-Speech Quality Index, with task completion rates of 68.6% on τ-Voice and 35.1% on Sierra's τ-Voice-banking benchmark. GPT-Live-1, paired with GPT-6 Astra at medium reasoning effort, scored 86.2% Pass@1 on Tau3's spoken customer-service evaluation spanning airline, retail, and telecom domains. On Full Duplex Bench, it beat GPT-Realtime-2.1 by 30 percentage points. Google cautions against treating that split as a direct comparison, noting that Google and OpenAI used different tests and setups. A Google spokesperson tells The New Stack that today's models center on providing developers and enterprises with tools to build voice agents, while ChatGPT and Claude voice mode are full products rather than models, making the comparison not apples-to-apples.
Per-minute costs diverge sharply between the two approaches. Standard Gemini 3.8 Live carries Gemini Live API rates of $0.005 per minute of audio input and $0.018 per minute of output, according to the report. Extended Thinking adds reasoning tokens, with additional charges for inputs like live video and documents. GPT-Live-1 costs $0.05 per voice minute for the front-end voice layer alone, with the backend reasoning model, function calls, and external agent runs all billed separately. Both approaches face the same challenge when a user interrupts or changes their mind halfway through a request, leaving background work running that may no longer be needed. In Gemini, that work stays within the same session, although developers have less visibility into exactly when a tool call stops. OpenAI leaves more of that cleanup to developers, who have to cancel pending jobs and ensure an outdated answer doesn't find its way back into the conversation. Google keeps speech, reasoning, and tool execution inside one session, cutting down on middleware but tying developers more closely to its runtime, while OpenAI requires more orchestration but gives developers more control over the models and tools running behind the voice layer. The architectural split reflects a broader divide: whether to embed complexity inside the model or push it out to the application layer, with cost and control hanging in the balance. For enterprises building voice agents at scale, the choice between a tightly integrated session and a loosely coupled stack will likely hinge less on benchmark scores and more on how much control they're willing to trade for simplicity.