Story perspectives

OpenAI Debuts GPT-Live Voice Model, Scores 76% Preference

7/10/2026

Summary

OpenAI Launches GPT-Live

OpenAI launched GPT-Live, a full-duplex voice model for ChatGPT on July 8, rolling out on iOS, Android and the web.
GPT-Live-1, the voice for Go, Plus and Pro plans, received a 75.7 % preference rating, while GPT-Live-1 mini for Free users earned 69.2 %.
Safety updates add real-time interventions, teen protections and blocks on mimicking real voices.
OpenAI will later offer GPT-Live via API, with a sign-up form now open to developers and enterprises.

Summary

OpenAI Limits Voice

OpenAI says voice recordings are kept for 30 days, can be deleted, and the new mode automatically opts users out of AI training.
Critics including Meredith Whittaker, president of Signal, warn that human-like assistants may create unhealthy attachment, citing AI anthropomorphism lawsuits.
OpenAI notes over 150 million weekly users of ChatGPT voice or dictation and that the mode listens continuously until disabled, with a wake-word to pause.

Summary

OpenAI Sparks Controversy

OpenAI launched GPT-Live as a voice-operating system with visual cards, future phone-call tasks and a developer API.
Early user feedback describes GPT-Live as “annoying” because its frequent “mhmm” and “got it” interjections feel intrusive.
Google’s Gemini Live, Nvidia’s PersonaPlex, Kyutai’s Moshi and academic prototypes already provide full-duplex voice, raising competition.
A new regulatory framework will study emotional dependency on voice assistants, targeting GPT-Live’s psychological impact.

Summary

OpenAI Unveils GPT-5.6

OpenAI will launch GPT-5.6 publicly on July 9 with three models: Sol, Terra and Luna.
Sol costs $5/$30 per-million input/output tokens, Terra $2.50/$15, and Luna $1/$6.
OpenAI added a prompt-caching mechanism to reduce token waste and lower costs.
Investors view the voice launch as the core growth driver ahead of an IPO.
The firm mentioned possible AI earbuds but gave no hardware details.

History

Historical Context

Siri began as a spin-out from SRI International’s AI Center, built on DARPA’s CALO project and co-founded by Dag Kittlaus, Tom Gruber and Adam Cheyer. Apple bought the startup in April 2010 and launched the integrated voice assistant on the iPhone 4S on October 4 2011, sparking consumer adoption and early privacy debates that echo today’s AI regulatory scrutiny.

1 of 5

OpenAI Launches GPT-Live

OpenAI launched GPT-Live, a full-duplex voice model for ChatGPT on July 8, rolling out on iOS, Android and the web.
GPT-Live-1, the voice for Go, Plus and Pro plans, received a 75.7 % preference rating, while GPT-Live-1 mini for Free users earned 69.2 %.
Safety updates add real-time interventions, teen protections and blocks on mimicking real voices.
OpenAI will later offer GPT-Live via API, with a sign-up form now open to developers and enterprises.

1 / 5