OpenAI Rolls Out GPT‑5.4
OpenAI deployed GPT‑5.4 across ChatGPT, Codex and its API and is shifting how it charges for developer tools, including a new Codex rate card and a reported $100 Pro tier for heavier users. The rollout touches consumer chat, coding tools and the API footprint while the published Codex pricing details show clearer product tiering for power users and businesses. (chatgptimagegenerator.org) (help.openai.com) (economictimes.indiatimes.com)
OpenAI is turning one model release into three separate product changes at once: GPT‑5.4 is now in ChatGPT, in the application programming interface that outside developers plug into, and in Codex, the coding agent OpenAI sells to programmers. The model itself is aimed at office work more than chatbot small talk. OpenAI says GPT‑5.4 combines reasoning, coding, and agent workflows in one system and is its “most capable and efficient” model for professional work. Inside ChatGPT, the new version shows up as GPT‑5.4 Thinking, which can lay out a plan before it finishes the answer. OpenAI says that lets users steer long responses midstream instead of waiting for a full draft and starting over. Inside Codex and the application programming interface, GPT‑5.4 adds native computer use, which means the model can operate software tools the way a human worker clicks through apps. OpenAI also says it can handle up to 1,050,000 tokens of context, which is roughly the difference between reading a memo and carrying a whole filing cabinet into one session. OpenAI is also using the rollout to redraw its price map for developers. On April 2, 2026, it changed Codex pricing for ChatGPT Business and new ChatGPT Enterprise customers from per-message billing to token-based billing, which charges for the amount of text going in and out instead of counting each prompt like a parking meter. The new Codex rate card makes the tiers much easier to compare. For GPT‑5.4 in Codex, OpenAI lists 62.50 credits per million input tokens, 6.250 credits per million cached input tokens, and 375 credits per million output tokens, while GPT‑5.4 Mini is far cheaper at 18.75, 1.875, and 113 credits respectively. That pricing shift matters because Codex is no longer being sold like a chat window with fuzzy limits. OpenAI’s help pages now describe Codex usage in terms closer to cloud computing, and the company says average Codex spend lands around $100 to $200 per developer per month depending on model choice, fast mode, and how many coding jobs are running at once. Then came the consumer side of the same strategy. On April 9, 2026, OpenAI announced a new $100 per month ChatGPT Pro tier that offers 5 times more Codex usage than Plus, while keeping access to the broader Pro feature set. OpenAI is even juicing the launch with a temporary bonus. The company said the new $100 Pro subscribers get up to 10 times Plus-level Codex usage through May 31, 2026, and it described the plan as built for “real projects” rather than occasional experiments. Put together, the move says OpenAI wants one flagship model to cover chat, coding, and agent work, while charging light users, heavy solo coders, and company workspaces in different ways. The model launch is the visible part, but the more durable change is that OpenAI is starting to price artificial intelligence less like a subscription box and more like a utility bill.