OpenAI Model Changes

OpenAI has pulled several older models out of public ChatGPT and replaced them with a narrower, more controlled lineup. (help.openai.com) OpenAI’s help center says GPT-4o, GPT-4.1, GPT-4.1 mini, OpenAI o4-mini, and GPT-5 Instant and Thinking were retired from ChatGPT on February 13, 2026. The company said API access to those models did not change. (help.openai.com) A second round followed on March 11, 2026, when OpenAI removed GPT-5.1 Instant, GPT-5.1 Thinking, and GPT-5.1 Pro from ChatGPT. OpenAI said existing GPT-5.1 conversations would continue on GPT-5.3 Instant, GPT-5.4 Thinking, or GPT-5.4 Pro. (help.openai.com) OpenAI is also rolling out GPT-5.3 Instant Mini in ChatGPT. The company said the new mini model feels “more natural in conversation” and outperforms GPT-5 Instant Mini across a range of use cases. (help.openai.com) The split is now clearer: ChatGPT is getting fewer public model choices, while the application programming interface, or API, still exposes older models for developers building their own products. OpenAI’s ChatGPT help pages repeat that the retirements apply to ChatGPT and that API availability remains in place. (help.openai.com) That matters most in cybersecurity, where OpenAI is no longer treating its strongest defensive models like general-purpose chat options. Last week, the company said it was expanding its Trusted Access for Cyber program and introducing GPT-5.4-Cyber, a version of GPT-5.4 fine-tuned for defensive security work. (openai.com) OpenAI said Trusted Access for Cyber is being scaled to “thousands” of verified individual defenders and “hundreds” of teams, instead of broad public release. In a February announcement, the company described the program as an identity- and trust-based framework meant to place stronger cyber capabilities “in the right hands.” (openai.com, openai.com) OpenAI said companies already participating in those cyber efforts include Cisco, Cloudflare, CrowdStrike, Oracle, Palo Alto Networks, and Zscaler, alongside banks including Bank of America, Citi, Goldman Sachs, JPMorgan Chase, and Morgan Stanley. The company also said it provided GPT-5.4-Cyber access to the U.S. Center for AI Standards and Innovation and the U.K. AI Security Institute for evaluations. (openai.com) For regular ChatGPT users, the immediate effect is simpler: fewer visible model names and more automatic migration to newer defaults. For OpenAI, the change leaves consumer chat on a shorter leash while keeping developer access and high-end cyber use behind separate gates. (help.openai.com, help.openai.com, openai.com)