OpenAI rolls out GPT-5.4 in ChatGPT as it updates model lineup

OpenAI said on March 5, 2026, that GPT-5.4 is now available in ChatGPT for paid users, the latest update to the company’s flagship chatbot as it continues to narrow and reorganize the model choices inside the app. The release matters because ChatGPT remains OpenAI’s most visible consumer product, and the company is using the rollout to push more users toward a simplified set of reasoning options while retiring older models.

The announcement comes amid a steady cadence of ChatGPT changes in March and April, including model retirements, interface updates and new fallback behavior for users who hit usage limits. OpenAI has been trying to make the product easier to navigate while keeping access to its newest reasoning systems available across different subscription tiers.

GPT-5.4 arrives in ChatGPT for paid users

In a post on its website, OpenAI said GPT-5.4 Thinking is available in ChatGPT for Plus, Team and Pro users, replacing GPT-5.2 Thinking. The company said GPT-5.2 Thinking will remain available for three months for paid users in the model picker under the Legacy Models section before being retired on June 5, 2026.

OpenAI said Enterprise and Edu customers can enable early access through admin settings, while GPT-5.4 Pro is available to Pro and Enterprise plans. The company described GPT-5.4 as its first mainline reasoning model to incorporate capabilities from GPT-5.3-Codex, its coding-focused model.

A narrower model picker

The GPT-5.4 rollout follows a March 17 update that simplified the ChatGPT model picker into a smaller set of options. According to OpenAI’s release notes, users on paid plans now see choices such as Instant, Thinking and Pro, with an Auto option available for switching between modes. OpenAI said the change is meant to make it easier to choose the right level of reasoning for a task.

OpenAI also said on March 16 that it updated GPT-5.3 Instant to improve follow-up tone and reduce teaser-style phrasing. A day later, the company said it was sunsetting the preset “Nerdy” base style, another sign that it is continuing to tune both the mechanics and tone of ChatGPT’s responses.

Older models are being retired

The company has also been removing older ChatGPT models from the product. OpenAI’s release notes say GPT-5.1 models were retired on March 11, 2026, and that GPT-4o, GPT-4.1, GPT-4.1 mini and OpenAI o4-mini were retired from ChatGPT on February 13, 2026, alongside GPT-5 Instant and GPT-5 Thinking.

Those changes reflect a broader effort to reduce the number of models users must sort through. OpenAI has said the goal is to make ChatGPT simpler while preserving access to more advanced reasoning for users who need it.

What to Watch

Attention now turns to whether OpenAI extends GPT-5.4 more broadly across ChatGPT tiers and whether it keeps trimming legacy models from the picker. The next test will be whether the simplified structure makes the product easier to use without limiting access to the specialized tools that have become central to ChatGPT’s appeal.