The Circuitry
THE CIRCUITRYYour one-stop source for all tech news
HOMETODAYNEWSFEEDEVENTS
BOOKMARKS
RSS
© 2026 The Circuitry
About UsSourcesContactCorrectionsPrivacy
  • Today
  • Feed
  • Events
  • Saved
Scroll for more
Verification
VERIFIEDConfidence: HIGH
Source identified
Claims cross-referenced
No discrepancies found
Fact-check summary

The Verge, Seeking Alpha, OpenAI's help docs, and other outlets confirm OpenAI's GPT-Live-1 full-duplex voice upgrade rolling out to ChatGPT today.

Sourcing
2independent sources

via 9to5Mac

9to5Mac · track record
64Stories
100%Verified
1030d
All sources →
Home/Tech/OpenAI upgrades ChatGPT voice mode using new GPT-Live-1
VERIFIEDBy Xavier Rivera· ·1.5 min read

OpenAI upgrades ChatGPT voice mode using new GPT-Live-1

OpenAI has revamped ChatGPT voice mode by deploying the full-duplex GPT-Live-1 model that feels more conversational, cuts unnecessary interruptions, supplies contextual visuals, forwards tough queries to stronger text models, and supports live translation while rolling out across platforms with tiered access for subscribers.

Source:9to5Mac
Post
OpenAI upgrades ChatGPT voice mode using new GPT-Live-1
TL;DRAI · 60 sec read

OpenAI replaces ChatGPT's turn-based voice system with the full-duplex GPT-Live-1 model. It speaks and listens simultaneously, interrupts less, handles pauses, routes complex queries to top text models, adds visuals, and supports real-time translation. The upgrade reaches iOS, Android, and web for paid subscribers, with a mini version for free users.

OpenAI has revamped the voice capabilities in ChatGPT by introducing the GPT-Live-1 model, which the company describes as feeling more like “talking to another person.”

OpenAI replaces turn-based voice with full duplex GPT-Live-1. The updated system reportedly interrupts speakers less frequently and pauses to let users resume if they hesitate during talks. During a press briefing, OpenAI research lead Kundan Kumar described GPT-Live-1 as the company’s “smartest voice model” yet.
What it really means is that it can speak and listen at the same time.

ChatGPT had previously depended on an older, turn-based voice model. “This is a full duplex model,” OpenAI product lead Atty Eleti said. “What it really means is that it can speak and listen at the same time.”
From The CircuitryThe Feed — live briefs across tech, all day.See what’s happening →
GPT-Live-1 routes complex queries to top text models and adds visuals. The model will hand off requests to leading text systems such as GPT-5.5 whenever deeper reasoning or web searches become necessary. Conversations covering topics like the weather, stocks, and sports will now include AI-generated visuals that display related details.
The full-duplex approach enables capabilities such as live translation even as the speaker continues.
New capabilities include real-time translation during speech. The full-duplex approach enables capabilities such as live translation even as the speaker continues. OpenAI has added built-in safeguards that steer the model away from harmful responses.
Rollout covers iOS, Android, web with tiered access. GPT-Live-1 is rolling out across iOS, Android, and on the web. It will power ChatGPT Voice for Go, Plus, and Pro subscribers, while a smaller GPT-Live-1 mini model will be the default for free users.
Why this mattersAI · ~100 words

Tap a lens to see what this story means for you.

Reader-supported
DonateBuy me a coffee →Follow@thecircuitry_ →Follow@thecircuitry.to →

Reader-supported · Daily Brief

Daily brief at 7 AM ET. Top tech stories, every morning. Sourced and fact-checked.

HELP US IMPROVE
From The Circuitry

See what’s happening right now

The Feed runs all day — short, verified briefs the moment they break.

Open the Feed →
From The Circuitry

Follow @thecircuitry_

Every story we publish, as it happens. No noise between.

Follow on X ↗On Bluesky ↗

Reader-supported

The Circuitry is a passion project I've always wanted to build, and I love the work behind it.

Running it costs real money. APIs, hosting, time. To keep improving the site and growing this into something useful for everyone, those costs have to be covered.

Any contribution is appreciated. If not, no pressure. Thanks for reading.

Buy me a coffee
OpenAIChatGPTVoice Mode
More from9to5Mac
  • Apple accuses OpenAI of trade secret theft in new lawsuit

    Tech · 5d
  • Apple commits over $30 billion to Broadcom in major US chip pact

    Tech · 8d
  • Apple asks California court to freeze Epic Games case during Supreme Court review

    Tech · 9d
More inTech
  • TSMC profit surges 77% as Arizona spending swells by $100 billion

    Tech · 13m
  • China's CXMT Doubles IPO Target to $8.55 Billion

    Tech · 16h
  • CISA Adds KNX Protocol CVE-2023-4346 to KEV Catalog

    Tech · 19h
SupportThe Work

The Circuitry is reader-supported. If you find the daily brief useful, you can buy me a coffee to keep it going.

Buy a coffee →
SubscribeCircuitry Brief

Daily brief at 7 AM ET. Top tech stories, every morning.

MORE IN TECH

TSMC profit surges 77% as Arizona spending swells by $100 billion

TSMC posted a 77.4% year-on-year profit increase to NT$706.56 billion and raised its capital spending outlook while announcing another $100 billion for Arizona fabs. The results underscore sustained AI demand that now dominates 66% of its platform revenue.

China's CXMT Doubles IPO Target to $8.55 Billion

ChangXin Memory Technologies expects to raise 57.9 billion yuan ($8.55 billion) in its Shanghai STAR Market IPO after doubling its original target. The world's fourth-largest DRAM chipmaker will list on July 27, advancing China's push for semiconductor self-reliance.

CISA Adds KNX Protocol CVE-2023-4346 to KEV Catalog

CISA added CVE-2023-4346 affecting KNX Association KNX Protocol Connection Authorization Option 1 to its Known Exploited Vulnerabilities catalog on 2026-07-15. Federal agencies must finish remediation by 2026-07-29. The overly restrictive account lockout mechanism could let attackers purge devices and set a BCU key when extra security options remain disabled, so organizations must apply vendor mitigations under BOD 26-04.