In March 2024, for instance, Altman mentioned in an interview with Lex Friedman that the corporate was getting ready the launch of a significant AI mannequin that yr (what seemingly grew to become GPT-4o or “o1”), however the last title was nonetheless up within the air. “I believe earlier than we discuss a GPT-5-like mannequin referred to as that, or not referred to as that, or just a little bit worse or just a little bit higher than what you’d anticipate from a GPT-5,” Altman mentioned to Friedman on the time, “I believe we’ve got a variety of different essential issues to launch first.”
It could be tempting to say that OpenAI might really feel a sure threshold of efficiency has been met to lastly label one in all its new AI fashions “GPT-5,” however the transfer apparently comes out of necessity. As talked about above, GPT-5 will primarily be a branding transfer that may consolidate many options into one unified interface.
“We need to do a greater job of sharing our meant roadmap, and a significantly better job simplifying our product choices,” Altman wrote in a submit on X. “We understand how difficult our mannequin and product choices have gotten.”
At present, if a person logs in to ChatGPT with a Professional account, they will select between an array of no less than 10 AI fashions, together with GPT-4o, GPT-4o with search, GPT-4o with deep analysis, GPT-4o with scheduled duties, o1, o3-mini, o3-mini-high, o1 professional mode, GPT-4o mini, and GPT-4. There’s additionally superior voice mode. There are two courses of AI fashions: what you may name “typical” LLMs (like GPT-4o) and simulated reasoning fashions (like o1 and o3-mini).
“A prime objective for us is to unify o-series fashions and GPT-series fashions by creating methods that may use all our instruments, know when to assume for a very long time or not, and usually be helpful for a really big selection of duties,” Altman wrote.
