SAN FRANCISCO, California: OpenAI has unveiled GPT-5, the latest version of the technology behind ChatGPT, promising significant gains in enterprise uses such as software development, finance, and health-related queries. The model is now available to all 700 million ChatGPT users.

The launch comes at a time of surging investment in AI, with Microsoft-backed OpenAI and rivals like Alphabet, Meta, and Amazon collectively set to spend nearly US$400 billion this fiscal year on AI infrastructure. The question is whether enterprise adoption will justify those costs.

“So far, business spending on AI has been pretty weak, while consumer spending … has been fairly robust because people love to chat with ChatGPT,” said economics writer Noah Smith. “But … it isn’t going to be nearly enough to justify all the money … spent on AI data centers.”

CEO Sam Altman called GPT-5 “the first time … one of our mainline models has felt like you can ask a legitimate expert, a PhD-level expert, anything.” He highlighted its “software on demand” abilities, with demos showing the model creating entire working programs from text prompts—known as “vibe coding.”

Two early reviewers told Reuters GPT-5 is strong in coding and solving science and math problems, though they felt the leap from GPT-4 was smaller than previous jumps. Altman noted GPT-5 still cannot learn on its own, a key step toward human-level capability.

The model incorporates “test-time compute,” which lets it spend more processing power on hard questions—technology now available to the public for the first time.

GPT-5’s release follows scaling challenges. After GPT-4’s 2023 debut, which outperformed GPT-3.5 by passing the bar exam in the top 10 percent versus the bottom 10 percent, OpenAI sought further gains by adding compute power and data. But it hit a “data wall,” as internet-scale human text datasets ran out. Training large models also risks hardware failures late in multi-month runs.

OpenAI’s $500 billion valuation goal, up from $300 billion, and reports of $100 million signing bonuses for top AI researchers underscore the high stakes. Altman insists global infrastructure investment is still insufficient: “We need to build a lot more … to have AI locally available in all these markets.”