OpenAI Will Stop Supporting Several Legacy ChatGPT Models in 2026
The company OpenAI has officially announced that starting from February 13, 2026, it will cease support for a series of previous versions of its artificial intelligence assistant. This measure affects models such as GPT‑4o, GPT‑4.1, GPT‑4.1 mini, and o4‑mini, which the company considers obsolete following the launch of GPT‑5 and its subsequent improvements. 🗓️
The End of a Cycle for Previous Models
OpenAI argues that these models have completed their useful lifecycle. Although access to GPT‑4o was temporarily restored in the past at the request of users who valued its conversational style, the decision is now final. The company assures that everything these systems could offer has already been incorporated into the main ChatGPT service.
Models Affected by the Discontinuation:- GPT‑4o: A model known for its particular conversational tone.
- GPT‑4.1 and GPT‑4.1 mini: Intermediate versions prior to GPT‑5.
- o4‑mini: A compact variant from a previous generation.
Maintaining legacy models active consumes processing capacity that we can dedicate to innovation.
Why OpenAI Is Making This Decision
The central reason is a resource strategy. OpenAI prefers to concentrate its efforts on developing and improving the latest technology, rather than maintaining parallel code lines that no longer represent its technological core. Key capabilities, such as the conversational style of those assistants, have been integrated or surpassed in the current versions.
Key Objectives After the Withdrawal:- Free up computing power for innovation in new projects.
- Avoid fragmenting development across multiple branches.
- Simplify infrastructure and maintenance.
The Future Path of ChatGPT
OpenAI communicates that its focus is now on continuously improving ChatGPT. They plan to refine aspects such as creativity and the control that users have over the generated responses. The goal is to evolve the main platform to create more useful and adaptable assistants, without the distractions of past technologies. Some users who missed the tone of GPT‑4o will have to adapt to the fact that new models not only respond but also actively suggest what to ask next. 🚀