ChatGPT Pro price cut
OpenAI appears to have halved ChatGPT Pro pricing to $100 per month and quietly upgraded a fallback model used during rate limits, lowering the cost of experimenting with capable AI tooling. Cheaper baseline access like this makes it easier for teams to prototype features but raises the bar for vertical platforms to justify a premium. (techradar.com)
OpenAI now has two ChatGPT Pro plans instead of one: a new $100 monthly tier and the older $200 tier, and both sit above the $20 Plus plan in the same settings menu. The company’s own Help Center says the $100 plan was added, not swapped in for the old one. (help.openai.com) The split is mostly about how much work you can push through, not which buttons you can see. OpenAI says both Pro plans include the same core features, while the $100 plan gives 5 times the usage of Plus and the $200 plan gives 20 times the usage of Plus. (help.openai.com) The product this seems built around is Codex, OpenAI’s coding agent inside ChatGPT. In OpenAI’s April 9 community post, the company said the new $100 tier offers 5 times more Codex usage than Plus and is aimed at “longer, high-effort Codex sessions.” (community.openai.com) OpenAI changed Codex pricing one week earlier, on April 2, 2026, from per-message billing to token-based billing for Plus, Pro, Business, and new Enterprise plans. That means usage now works more like a cloud meter, where long prompts, long outputs, and repeated runs show up directly in the billable credit count. (help.openai.com) OpenAI’s own rate card says Codex now costs roughly $100 to $200 per developer per month on average, with big swings based on model choice, fast mode, and how many automations are running. A cheaper Pro tier landing right after that pricing change gives individual developers and small teams a lower monthly step-up before they jump to the $200 plan. (help.openai.com) The other quiet change is what happens when a user hits a model cap. OpenAI’s release notes on April 9 said ChatGPT now falls back from GPT-5.3 Instant to GPT-5.3 Instant Mini, and that fallback model does not appear in the model picker because it only shows up after rate limits are reached. (help.openai.com) That sounds small, but fallback models are the spare tire of a chatbot subscription. If the spare tire gets better, the product feels less broken at the exact moment a paying user would normally notice the limits most. (help.openai.com) OpenAI has been building more of these layered fallbacks across ChatGPT this year. In March, the company said GPT-5.4 mini would serve as the rate-limit fallback for GPT-5.4 Thinking, which shows a pattern: keep users inside the product even after they hit the premium lane. (help.openai.com) The timing also lines up with a broader push toward more capable work models. OpenAI introduced GPT-5.4 on March 5 as its flagship model for professional work, with stronger coding, tool use, and up to 1 million tokens of context in Codex and the application programming interface. (openai.com) So the story is not just “price cut.” OpenAI is turning ChatGPT into a staircase of usage bands, where $20 gets you in, $100 covers real weekly project work, and $200 is for people running parallel heavy workflows without stopping. (help.openai.com) That changes the math for every startup selling an artificial intelligence wrapper on top of general-purpose models. When the base subscription gets cheaper and the fallback gets better, the easy premium disappears first. (help.openai.com)
