OpenAI prices Codex Pro

OpenAI rolled out a $100/month Pro plan for its Codex coding product and published a tiered rate card that clarifies how credits and quotas work across consumer and enterprise plans. This shifts coding access from a loose “model sandbox” into a metered product where usage, routing, and overages are practical concerns for builders of embedded copilots. (help.openai.com) (economictimes.indiatimes.com)

OpenAI just turned its coding product into something that looks a lot more like a cloud bill. On April 2, 2026, it replaced Codex’s old per-message estimates with token-based pricing and started selling a new Pro tier from $100 a month. (help.openai.com 1) (help.openai.com 2) A token is the small unit these systems count when they read and write text, like a taxi meter counting distance instead of trips. OpenAI’s new rate card now charges Codex by input tokens, cached input tokens, and output tokens, so the size of the codebase and the size of the answer both change the bill. (help.openai.com) The cheapest part is cached input, which is reused context the system has already seen once. In OpenAI’s table, 1 million cached input tokens on GPT-5.4 cost 6.25 credits, while 1 million fresh input tokens cost 62.5 credits and 1 million output tokens cost 375 credits. (help.openai.com) That means a long coding session is no longer “one more prompt.” A task that makes Codex reread a large repository, write a lot of new code, or run in fast mode can burn through credits much faster than a short local edit. (help.openai.com) (developers.openai.com) The new consumer ladder now has a clearer shape. ChatGPT Plus stays at $20 a month, while Pro starts at $100 a month and offers 5 times the Plus usage, and OpenAI says the existing $200 Pro option remains for heavier users with 20 times the Plus usage. (help.openai.com) (developers.openai.com) OpenAI is also using Codex to split “I use ChatGPT at work” from “I need an engineering agent all day.” Its April 2 announcement says ChatGPT Business and ChatGPT Enterprise now have standard seats with included limits and Codex-only seats with no rate limits, where usage is billed entirely on token consumption. (openai.com) (help.openai.com) That seat split changes who can buy in. A company can now give one engineer a Codex-only seat for a narrow pilot instead of buying a full ChatGPT seat for everyone on the team. (openai.com) (help.openai.com) OpenAI is also trying to make those pilots easier to start. It said eligible ChatGPT Business workspaces can get $100 in credits for each new Codex-only member, up to $500 per team, and it cut the annual ChatGPT Business price from $25 to $20 per seat. (openai.com) The company’s own pricing page now reads less like a chatbot menu and more like an infrastructure product sheet. Plus includes Codex on the web, in the command line interface, in the integrated development environment extension, and on iPhone, while the application programming interface key option is positioned for shared automation and continuous integration with token-based billing. (developers.openai.com) OpenAI says average Codex usage lands around $100 to $200 per developer per month, but it also says that number swings with model choice, automations, fast mode, and how many instances a user runs. That is a very different promise from “unlimited coding help,” because now the expensive part is not signing up but keeping a busy agent fed with context. (help.openai.com) The competitive angle is straightforward. Economic Times reported that OpenAI is pitching the $100 Pro plan against Anthropic’s Claude Max, but the bigger shift is that coding assistants are starting to look like metered coworkers: cheap when they review a few files, expensive when they sit on every pull request, every repository, and every automation loop. (economictimes.indiatimes.com) (help.openai.com)

Get your own daily briefing

Scout delivers personalized news, insights, and conversations tailored to your role and industry.

Download on the App Store

Shared from Scout - Be the smartest in the room.