Particle.news
Download on the App Store
19 ARTICLES
·
3w ago

OpenAI Releases GPT‑5.6 and GPT‑Live After U.S. Security Review

The company opened public access following government tests, creating a test case for whether voluntary federal previews of frontier models become routine.

Image
OpenAI presentó GPT-Live, para hablar con la IA como si fuera una persona; incluso es posible activar la IA con un comando de voz al estilo Siri o Gemini
Image
Stanford puso a disposición del público una clase gratuita que explica cómo funcionan por dentro modelos de IA como ChatGPT y Claude.

Overview

  • OpenAI released the GPT‑5.6 family on Thursday, July 9, expanding preview access to three models called Sol, Terra and Luna that target high‑end, mid‑tier and low‑cost uses respectively.
  • The U.S. Department of Commerce and other federal experts conducted additional tests during a restricted preview before wider launch, and reporting says those reviews cleared a broader rollout.
  • OpenAI also began a global, staged rollout of GPT‑Live, a full‑duplex voice system that listens and speaks simultaneously and delegates complex or long tasks to a more powerful background model (currently GPT‑5.5) to keep conversations fluid.
  • The company published a technical safety report and said GPT‑Live includes real‑time monitoring, intervention rules for risky content, and tiered availability with GPT‑Live‑1 for paid plans and GPT‑Live‑1 Mini for free users.
  • The releases sharpen a trade‑off between fast commercial deployment and national‑security oversight, with OpenAI cooperating with U.S. reviewers while warning that routine government pre‑release gatekeeping could limit access for developers and international partners.