OpenAI’s $100 Codex Tier
OpenAI introduced a $100/month ChatGPT Pro tier aimed at heavy Codex users, offering roughly five times the Codex usage of the $20 Plus plan and signaling that AI coding is being sold as an operational budget line. Reports also claim a GPT‑5.4 rollout with experimental one‑million‑token context support in Codex, though that model claim is less verified. Together the pricing and model signals imply vendors expect frequent, workflow‑native AI use rather than occasional experimentation. (techcrunch.com) (chatgptimagegenerator.org)
OpenAI just put a new price tag on heavy artificial intelligence coding: $100 a month. The new ChatGPT Pro tier sits between the $20 Plus plan and the older $200 Pro plan, and OpenAI says it is built for people using Codex every day instead of occasionally. (community.openai.com) (techcrunch.com) Codex is OpenAI’s coding agent inside ChatGPT. OpenAI’s help pages describe it as a tool that can write code, review code, and complete software tasks either in local tools like a terminal and integrated development environment or in the cloud. (help.openai.com) The new $100 plan is mostly about usage, not new basic features. OpenAI says the $100 tier gives 5 times more Codex usage than Plus, while the $200 Pro plan remains the highest tier at 20 times Plus usage. (help.openai.com) (community.openai.com) OpenAI also says the $100 plan is for “longer, high-effort Codex sessions” and parallel projects. That language sounds less like a chatbot sold for curiosity and more like a work tool sold for hours of output, the way a company budgets for cloud storage or software seats. (community.openai.com) The timing matters because OpenAI had a big gap in the lineup. TechCrunch reported that ChatGPT now had free, $8 Go, $20 Plus, and $200 Pro options, and the new $100 tier fills the jump between casual use and the most expensive individual plan. (techcrunch.com) OpenAI is also tightening the link between subscription plans and metered coding work. Its Codex pricing page says Plus and Pro users who hit their limits can buy extra credits instead of changing plans, which makes the product look more like a utility bill than an all-you-can-use subscription. (developers.openai.com) There is a second piece to this story: the model underneath Codex has also moved up. OpenAI said on March 5, 2026 that GPT-5.4 was rolling out in ChatGPT, the application programming interface, and Codex, and described it as its main model for professional work. (openai.com) (developers.openai.com) That GPT-5.4 release is where the “one million token context” claim comes from. OpenAI’s own product post says GPT-5.4 supports up to 1 million tokens of context, and its model docs list a 1.05 million token context window for GPT-5.4 and GPT-5.4 Pro. (openai.com) (developers.openai.com) A context window is the amount of text a model can keep in view at once. A million-token window means a coding agent can hold far larger codebases, logs, specifications, and test results in working memory before it has to summarize or drop details. (openai.com) (developers.openai.com) OpenAI’s Codex model page now tells users to start most tasks with GPT-5.4. That matters because the company is not treating advanced coding as a side feature anymore; it is making the default Codex experience depend on its newest general-purpose model. (developers.openai.com) Put those two moves together and the pitch becomes clear in concrete terms: pay more each month, get more coding sessions, and if that still is not enough, buy more credits. That is a product built for repeated, workflow-level use, not for a developer asking for a few snippets after lunch. (help.openai.com) (developers.openai.com)
