OpenAI has unveiled its newest family of artificial intelligence models, GPT-5.6, designating it the 'preferred model' for Microsoft's Copilot 365. The release comes amid reports that the partnership between the two tech giants may be shifting, though neither company has confirmed a breakup.
The launch marks a significant step in frontier intelligence, with the model offering improved scalability according to OpenAI. The New York Times describes GPT-5.6 as the company's most powerful AI model yet, while The Guardian notes the release follows a delay due to White House cybersecurity concerns.
CEO Sam Altman told CNBC that the model is 54% more token efficient on agentic coding. Bloomberg reports the company also unveiled a ChatGPT Work Agent designed to autonomously handle tasks for hours, and Reuters notes the release of a long-awaited 'super app' as competition with Anthropic intensifies.
The implications for enterprise users are substantial, as Copilot 365 integration could reshape how businesses leverage AI for productivity. Meanwhile, TechCrunch reports OpenAI is shutting down Atlas, its standalone desktop browser, signaling a consolidation strategy focused on core AI products.
Critics argue that OpenAI's rapid iteration pace raises reliability and safety concerns, especially after the cybersecurity-related delay. The heavy reliance on Microsoft's ecosystem also risks vendor lock-in for customers.