Image: Matheus Bertelli, Pexels
on the roadmap for ChatGPT. In the post, he explained they’ve halted their upcoming o3 reason model to focus on a streamlined but monolithic GPT-5.
We want AI to “just work” for you. We realize how complex our model and product offerings are.
As much as you hate the model selector, we want to go back to magic unified intelligent.
As our last non-chain of thought model, we will ship GPT-4.5. This is the model that we internally called Orion.
…
We will release GPT-5 in both ChatGPT as well as our API. This system integrates many of our technologies, including o3. We will no long ship o3 standalone.”
OpenAI released in January a lighter version called o3 miniwhich is especially well-suited for STEM subjects and coding. The plan was to release a broader version of o3 shortly afterward. The new plan is instead to launch GPT-4.5 next, which will be its last AI model that does not use chain-of-thought.
OpenAI plans to unify its o-series AI models with the GPT-series by creating a system that combines all tools and knows how to prioritize them for a specific task. This new system, GPT-5 will also include the upcoming model o3. GPT-4.5 is due to be released in the next few weeks, and GPT-5 will follow in a couple of months.
Altman says that the free version will have unlimited access GPT-5 on the standard intelligence level. ChatGPT Pro subscribers will have access to GPT-5 running at a higher level of intelligence, while ChatGPT Plus users will be able run it at a still higher level.
Further Reading: Microsoft gives ChatGPT’s $200/mo AI for Windows users for free
Originally published in our sister publication PC for Alla, this article was translated and localized.
