ChatGPT Pro $100 tier

OpenAI has added a new $100 a month ChatGPT plan, and the whole point is simple: some programmers were hitting the $20 plan’s coding limits long before they needed the $200 plan’s firehose. OpenAI announced the new tier on April 9, 2026, and said it is built for “longer, high-effort Codex sessions.” (community.openai.com) Codex is OpenAI’s cloud coding agent, which means you hand it software tasks and it works through them on remote machines instead of just spitting out one reply in a chat box. OpenAI introduced Codex in May 2025 and said it could handle multiple software engineering tasks in parallel. (openai.com) The new $100 tier sits between ChatGPT Plus at $20 a month and the older top-end Pro plan at $200 a month. OpenAI’s help page says the difference between the two Pro plans is mostly usage allowance, not a different feature set. (help.openai.com) On Codex, the $100 plan gets 5 times the usage of Plus, while the $200 plan gets 20 times the usage of Plus. OpenAI also says the $100 plan has a temporary launch bonus through May 31 that raises it to 10 times Plus for Codex use. (community.openai.com) Inside ChatGPT, OpenAI says the $100 plan still includes all Pro features, including the exclusive Pro model and unlimited access to its Instant and Thinking models. That means OpenAI is not selling a weaker toolset at $100; it is mostly selling a smaller bucket of heavy-duty usage than the $200 tier. (community.openai.com) (help.openai.com) There is another shift underneath this launch: Codex billing changed on April 2, 2026, from per-message pricing to pricing tied to application programming interface token usage. In plain English, OpenAI moved from charging by each back-and-forth to charging more like a meter that tracks how much model compute your coding job actually burns. (help.openai.com) That matters because coding agents do not all cost the same amount to run. A quick bug fix might take one short pass, while a big refactor can chew through long context windows, repeated tests, and parallel tasks across several projects. (openai.com) (developers.openai.com) OpenAI is also responding to a rival that has been gaining ground with programmers. TechCrunch and CNBC both reported that OpenAI is openly positioning the new $100 tier against Anthropic’s Claude coding products, especially as developers compare price bands before they commit their daily workflow to one assistant. (techcrunch.com) (cnbc.com) The pricing ladder is now clearer than it was a week ago: free for light use, $20 for regular users, $100 for heavy individual coders, and $200 for the people who want the highest limits OpenAI sells directly in ChatGPT. OpenAI’s help center says users can switch between the $100 and $200 Pro plans from Settings and have billing adjusted automatically. (help.openai.com) The real story is not that OpenAI found a new number between 20 and 200. It is that coding assistants are turning into metered infrastructure, and OpenAI now wants to charge more like a cloud provider that knows exactly which users run a bicycle, which users run a delivery van, and which users run a freight train. (developers.openai.com) (help.openai.com)