OpenAI has unveiled a new, more powerful family of artificial intelligence models, according to The New York Times. TechCrunch reports the launch centers on GPT-5.6, the newest addition to the company's lineup of models, while Forbes notes OpenAI is also introducing what it calls "GPT-Live" models.
Alongside the models, OpenAI is pushing further into AI "agents" — software meant to carry out tasks on a person's behalf rather than just answer questions. Ars Technica reports the company has rebranded its Codex tool, promising independent workflows that can run "for hours if needed," and framing it as something designed to do your work both for you and with you.
Reuters, in reporting by Deepa Seetharaman and Juby Babu, describes the release as a long-awaited "super app." According to Reuters, OpenAI showcased a new AI agent on Thursday aimed at helping white-collar workers tap the power of coding tools without the "sticker shock" of high costs. Reuters frames the launch against an intensifying rivalry with competitor Anthropic.
Taken together, the sources point to a coordinated push: a stronger core model, new "live" variants, and agent tools built to complete multi-step tasks with limited supervision.
Why it matters: If AI agents can reliably run complex work for hours at a stretch, the technology moves from a tool people prompt one question at a time toward software that can take on chunks of everyday office work — raising the stakes in the race between OpenAI and Anthropic to define how people actually use AI.