OpenAI launches new Pro tier

OpenAI just put a new $100-a-month step in the middle of its ChatGPT ladder, closing the gap between the $20 Plus plan and the $200 Pro plan it launched in December 2024. The new plan gives paying users a cheaper way to get newer models and heavier coding use without jumping straight to the top tier. (venturebeat.com) (openai.com) The basic change is simple: OpenAI is no longer treating “paid ChatGPT” like one bucket. Its current pricing pages already split customers across Free, Go, Plus, Business, and Enterprise, and the new $100 plan adds another capability band aimed at people who use ChatGPT more like a work tool than a casual chatbot. (openai.com) (msn.com) The model at the center of this move is GPT-5.4, which OpenAI says is its most capable model for professional work. In ChatGPT, OpenAI says GPT-5.4 Thinking can show an upfront plan while it works through a problem, and in the application programming interface it comes with a context window of about 1.05 million tokens, which is enough room for very large documents and codebases. (openai.com) (developers.openai.com) The second piece is Codex, OpenAI’s coding product, which works like an extra software teammate that reads code, writes code, and handles long technical tasks. OpenAI’s research pages say GPT-5.3-Codex was built for “long-horizon” technical work, and reporting on the new $100 plan says it comes with 5 times the Codex usage limit of Plus. (openai.com) (venturebeat.com) That matters because OpenAI has been building Codex into a much bigger product line than a simple chat add-on. On April 2, 2026, OpenAI’s help center said Business and Enterprise customers got two seat types, including a new Codex-only seat, which means the company is already separating “general ChatGPT access” from “developer workflow access” inside team accounts. (help.openai.com) The pricing now starts to look more like cloud software than a consumer subscription. OpenAI’s application programming interface page lists GPT-5.4 at $2.50 per 1 million input tokens and GPT-5.4 Pro at $30 per 1 million input tokens, so there is already a wide spread between standard and premium model access on the developer side. (openai.com) (developers.openai.com) The old problem was that ChatGPT’s consumer plans had a cliff in the middle. A user who outgrew Plus had to jump from $20 to $200, and that jump made sense for a power user or a company expense account but not for a freelance programmer, startup founder, or small team lead who just needed more coding capacity. (openai.com) (engadget.com) OpenAI is also making this move while reshuffling which models live inside ChatGPT. Its help pages say GPT-5 Instant and GPT-5 Thinking were retired from ChatGPT on February 13, 2026, while newer GPT-5.4 variants rolled into ChatGPT, Codex, and the application programming interface in March 2026. (help.openai.com) (community.openai.com) So the new $100 tier is not just a price cut between two old plans. It is OpenAI turning ChatGPT into a menu where model quality, coding capacity, and seat type can be mixed more precisely, which is the kind of packaging companies use when they want to capture everyone from solo developers to enterprise procurement teams. (openai.com) (help.openai.com) If this holds, the next budget conversation inside companies gets more granular. Instead of asking whether a whole team should have “paid ChatGPT,” managers can now decide who gets a $20 chat plan, who gets a $100 coding-heavy plan, and who still needs the $200 top tier for maximum performance on the hardest tasks. (openai.com) (venturebeat.com)