Particle.news

Download on the App Store

GPT-5 Enters Enterprise Stack as OpenAI Eases Rollout Missteps and Teases Memory-Centric GPT-6

OpenAI is prioritizing product fixes and long‑term memory in response to early user backlash.

ANKARA, TURKIYE - AUGUST 13: In this photo illustration, the logo of OpenAI logo is being displayed on a mobile phone screen in front of another screen displaying a robotic hand, in Ankara, Turkiye on August 13, 2025. (Photo by Ismail Aslandag/Anadolu via Getty Images)
Image
Sam Altman, OpenAI CEO listens as President Donald J Trump speaks about infrastructure and artificial intelligence to reporters with  Larry Ellison, chairman of Oracle Corporation and chief technology officer, Masayoshi Son, and SoftBank Group CEO, in the Roosevelt Room at the White House on Tuesday, Jan 21, 2025 in Washington, DC.
Image

Overview

  • Microsoft made GPT-5 generally available across Microsoft 365 Copilot, GitHub Copilot and Azure AI Foundry, with multiple variants and large context windows managed by a new model router.
  • Oracle embedded GPT-5 into Oracle Database, Fusion Cloud Applications and NetSuite to add natural‑language queries, code assistance and agent‑driven automation inside core workflows.
  • Sam Altman acknowledged the company "screwed up" the rollout by replacing GPT-4o as the default, and OpenAI restored 4o as an option for paying subscribers while it tunes tone and behavior.
  • Independent tests report higher coding scores and fewer hallucinations versus prior models, and OpenAI leaders and product managers highlight design changes aimed at safer, more grounded answers.
  • Altman says GPT-6 will emphasize persistent memory and personalization, and he cautions that GPU capacity constraints could shape the release timeline and availability.