OpenAI retires GPT‑5 in ChatGPT
- OpenAI retired GPT-5 Instant and Thinking from ChatGPT on February 13, 2026, alongside GPT-4o, GPT-4.1, GPT-4.1 mini, and o4-mini. - Business, Enterprise, and Edu customers kept GPT-4o inside Custom GPTs only until April 3, while API access to retired models stayed unchanged. - OpenAI paired the removals with new rate cards, model limits, and GPT-Rosalind access for U.S. enterprise biology teams. (help.openai.com)
OpenAI removed GPT-5 Instant and GPT-5 Thinking from ChatGPT on February 13, 2026, along with GPT-4o, GPT-4.1, GPT-4.1 mini, and o4-mini. (help.openai.com) The company said the retired models are no longer available in ChatGPT, but API access to those models remains unchanged. Business, Enterprise, and Edu customers kept GPT-4o in Custom GPTs only until April 3, 2026. (help.openai.com 1) (help.openai.com 2) The replacement inside ChatGPT is a newer lineup: GPT-5.3 became the default model for logged-in users, and GPT-5.5 began rolling out to Plus, Pro, Business, and Enterprise users. GPT-5.5 Pro rolled out to Pro, Business, and Enterprise users. (help.openai.com) The change turned ChatGPT’s enterprise model menu into a pricing and controls question as much as a capability question. OpenAI’s current Business and Enterprise rate card charges 10 credits for GPT-5.5 Thinking and 50 credits for GPT-5.5 Pro, while GPT-5.3 Instant is listed as unlimited. (help.openai.com) OpenAI also published updated enterprise model limits and admin controls. In Enterprise and Edu workspaces, GPT-5.3 Instant has a 128K context window, GPT-5.5 Thinking has 196K, and admins can enable or disable access to additional models. (help.openai.com) The billing system changed at the same time. OpenAI said on April 2, 2026 that Business and Enterprise plans added a standard ChatGPT seat and a Codex-only seat under flexible pricing. (help.openai.com) Codex pricing moved away from per-message billing to token-based billing on April 2 for most plans, and on April 23 for all existing Enterprise plans, including Edu, Health, Gov, and ChatGPT for Teachers. (help.openai.com) OpenAI also introduced GPT-Rosalind, a life-sciences model for eligible U.S. customers with Enterprise agreements. During the research preview, those teams can use it in ChatGPT Enterprise, Codex, and the API for internal biology research workflows. (help.openai.com) GPT-Rosalind is not available for customer-facing products or external commercial applications, and OpenAI said individual researchers are not currently supported. The company said the model is aimed at target discovery, genomics interpretation, pathway analysis, and literature synthesis. (help.openai.com) The result is that OpenAI’s current ChatGPT lineup is narrower, but its enterprise catalog is more segmented. Customers now have to weigh credits, seat types, admin settings, and governance rules alongside model names. (help.openai.com 1) (help.openai.com 2)
