OpenAI’s $100 Pro Tier

OpenAI just split ChatGPT’s paid plans into a new middle tier: $20 for Plus, $100 for Pro, and $200 for the older top-end Pro plan that OpenAI says is still available. The new $100 option is aimed at people who burn through coding-agent usage faster than the $20 plan allows. (openai.com) (techcrunch.com) The product at the center of this is Codex, OpenAI’s coding agent inside ChatGPT. OpenAI describes it as a tool that can handle end-to-end software work like building features, refactoring code, and managing migrations instead of just answering one coding question at a time. (chatgpt.com) (help.openai.com) The pricing change is blunt: the new $100 plan includes 5 times as much Codex usage as ChatGPT Plus. OpenAI also says that, through May 31, 2026, that same plan gets up to 10 times the Codex usage of Plus as a launch promotion. (openai.com) (help.openai.com) OpenAI’s own help page now frames the tiers by workload. Plus is for people using advanced tools through the week, while Pro at $200 is for “heavy lifting” and offers 20 times the limits of Plus; the new $100 plan lands between them for longer coding sessions without jumping all the way to $200. (help.openai.com) (techcrunch.com) That tells you what changed inside ChatGPT over the past year: coding has gone from a side feature to a metered workload. OpenAI’s developer pricing page says Codex is bundled into ChatGPT plans, but users who hit limits can now buy extra credits, which makes the product look less like a flat subscription and more like cloud software with overage spending. (developers.openai.com) (help.openai.com) OpenAI also changed how some business Codex pricing works on April 2, 2026. Its rate card says new ChatGPT Business and new ChatGPT Enterprise customers are being aligned to application programming interface token usage instead of per-message pricing, which is the same kind of meter companies already use for cloud computing and application programming interfaces. (help.openai.com) The competitive target is obvious. TechCrunch and VentureBeat both describe the move as a bid to keep developers from drifting to Anthropic’s Claude Code and Google’s coding tools, especially among “vibe coding” users who want an agent to keep working across long sessions instead of stopping at a cap. (techcrunch.com) (venturebeat.com) There is also a quiet budgeting story here. A $100 seat is expensive enough that a team lead can no longer wave it through as a casual perk, but cheap enough that a manager can justify it for one engineer who uses Codex every day and would otherwise keep hitting the $20 ceiling. That is an inference from OpenAI’s new tier structure and credit system, not a line the company states directly. (openai.com) (developers.openai.com) So the headline is not just that ChatGPT got a new price. It is that OpenAI now sells coding help in three layers: casual at $20, professional at $100, and near-continuous at $200, with extra credits on top when usage spikes. (help.openai.com) (developers.openai.com)