OpenAI product line major adjustment: GPT-5 to be released within a few months, basic version free for unlimited use

Image source: Generated by Unbounded AI

On February 13, OpenAI CEO Sam Altman announced today through the social media platform X that the company will completely adjust its artificial intelligence product roadmap, cancel the originally planned independent release of the "o3" model, and instead launch the integrated GPT-5 with multiple technologies.

Note: GPT-4.5 and GPT-5 will be released in several weeks and several months respectively

According to Ultraman, OpenAI's adjusted roadmap will proceed in two steps: releasing GPT-4.5 (code-named Orion) as a transitional product in the coming weeks, which is considered internally as the "last non-chain thinking model" with limited improvements, but still needs to undertake the task of bridging the technical gap; the real game-changer will be GPT-5, which will be released in a few months.

It is understood that GPT-5 will end the split state of language model and reasoning model, automatically determine when rapid response is needed, when deep thinking is started through dynamic task processing mechanism, and integrate voice interaction, canvas creation, real-time search, and "deep research" functions at the same time.

Altman said that free users can use the basic version of GPT-5 unlimitedly, and the paid tier is based on a $20/month Plus subscription and a $200/month Pro subscription to unlock higher intelligence intensity and in-depth research times respectively - a strategy that not only lowers the threshold for experience, but also transforms technical advantages into layered monetization capabilities.

OpenAI's sudden adjustment of the product roadmap decision is directly derived from the 'model maze' that ChatGPT platform has fallen into due to rapid iteration over the past half year, with multiple model versions being released intensively, forcing users to repeatedly choose between 'GPT-4o regular Q&A', 'o3-mini high-speed inference', 'GPT-4o mini lightweight response', and even 'scheduled task version GPT-4o'.

Altman admits, "The out-of-control pace of product launches is the main cause of the complication, and we realized that users shouldn't be bothered by choosing a model – AI should be 'ready to use'. ”

Note: Roadmap of OpenAI's existing model

Currently, the ChatGPT Plus subscription interface is like a technical menu: the GPT-4o, which is the flagship for general scenarios, is labeled as "suitable for most problems", the "scheduled task version" that needs to be manually enabled allows for delayed response; the o1 and o3-mini, which focus on mathematics and logic respectively, emphasize "advanced reasoning" and "coding ability"; GPT-4 is still retained under the name of "classic model".

This fragmented experience even extends to the product naming system - from the "o-series" inference model to the "GPT series" language model, users need to understand the differences between the two technical routes in order to use them effectively.

Altman bluntly stated in X: "We hate model selectors as much as users, and the goal is to return to magically unified intelligence." ”