ChatGPT lineup simplified
OpenAI says it has retired several consumer‑facing GPT variants inside ChatGPT while keeping API access steady and preserving GPT‑4o for Business, Enterprise and Edu customers within custom GPTs. The change suggests OpenAI is curating the public ChatGPT surface while maintaining stability for managed organisational deployments. That divergence matters for teams building internal tools versus experimenting on the consumer platform. (help.openai.com)
OpenAI has started hiding a lot of the old model picker inside ChatGPT and replacing it with one default system called GPT-5.3, which the company says is now the standard experience for logged-in users. The same help page says GPT-5.3 Instant is still rolling out over the next few days, so some people may see the new setup before others. (help.openai.com) The old menu used to expose separate names like GPT-4o, GPT-4.1, GPT-4.1 mini, OpenAI o4-mini, and GPT-5 in Instant and Thinking versions. OpenAI says those models were retired from ChatGPT on February 13, 2026, even though application programming interface access stayed unchanged. (help.openai.com) That split is the key to the whole move. ChatGPT is the consumer app where people click a model name in a chat window, while the application programming interface is the developer pipe that lets companies call a model from their own software. (help.openai.com) OpenAI is not saying one model now does literally one thing. It says GPT-5.3 in ChatGPT is “a single auto-switching system” that pulls together the best parts of its models, which is closer to a car with an automatic transmission than a garage full of manual gearboxes. (help.openai.com) There is still one visible specialist model for harder work. OpenAI describes GPT-5.4 Thinking as its most capable reasoning model in ChatGPT and says it is designed for difficult tasks like spreadsheets, polished front-end code, slideshows, hard math, and research across many web sources. (help.openai.com) The company had been giving business customers a longer runway than ordinary ChatGPT users. OpenAI’s help center said ChatGPT Business, Enterprise, and Edu customers could keep GPT-4o inside custom GPTs until April 3, 2026, even after the public ChatGPT model list had already been cleaned up. (help.openai.com) That exception has now expired. OpenAI’s current Business, Enterprise, and Edu help pages all say GPT-4o was fully retired across all plans after April 3, 2026. (help.openai.com, help.openai.com) So the simplification is real, but it is not random. OpenAI removed old names from the public ChatGPT surface first, kept the application programming interface stable for builders, and gave managed workplace customers a short bridge period for custom GPTs before ending that too. (help.openai.com, help.openai.com) If you use ChatGPT like a search box or writing assistant, this mostly means fewer knobs. If you run internal tools through the application programming interface, the important sentence is still “API access remains unchanged,” which means the app got simpler without forcing the same change on software teams overnight. (help.openai.com, help.openai.com)
