OpenAI publishes Codex rate card

- OpenAI published a Codex rate card that shifts ChatGPT coding usage to token-based billing across Plus, Pro, Business, Enterprise, Edu, Health, Gov and Teachers plans. - The card prices GPT-5.5 Codex usage at 125 credits per million input tokens and 750 per million output tokens, replacing older per-message estimates. - The change follows April product updates that added ads to Free and Go in four countries while paid tiers stayed ad-free. (help.openai.com)

OpenAI has published a Codex rate card that turns ChatGPT coding usage into a metered, token-based product across consumer and enterprise plans. (help.openai.com) The help article says OpenAI switched Codex pricing from credits per message to credits per token on April 2, 2026 for Plus, Pro, Business and new Enterprise plans. It says the same change reached existing Enterprise customers, including Edu, Health, Gov and ChatGPT for Teachers, on April 23. (help.openai.com) The new table lists separate rates for input, cached input and output tokens, which means the bill changes with how much code or context a session reads, reuses and writes back. GPT-5.5 is priced at 125 credits per million input tokens, 12.50 cached-input credits and 750 output credits. (help.openai.com) Lower tiers on the same card include GPT-5.4 at 62.50 input credits and 375 output credits per million tokens, GPT-5.4 Mini at 18.75 and 113, and GPT-5.3 Codex at 43.75 and 350. The article also notes that Fast mode consumes credits at a higher rate on supported models. (help.openai.com) OpenAI tied the pricing shift to subscription changes earlier in April. ChatGPT release notes on April 9 introduced a new $100-a-month Pro plan, kept the $200 Pro option, and said the cheaper Pro tier would include, for a limited time, up to 10 times more Codex usage than Plus. (help.openai.com) Those same release notes said Plus would be rebalanced away from long, high-intensity coding bursts on a single day and toward more sessions across the week. In other words, OpenAI is not just selling access to a model name; it is selling different usage shapes. (help.openai.com) The billing language now lines up with how OpenAI is reorganizing ChatGPT itself. Model release notes say GPT-5.4 Mini became a fallback for GPT-5.4 Thinking on March 18, and GPT-5 Thinking Mini was set to be retired as a selectable model within 30 days. (help.openai.com) OpenAI had already retired GPT-5.1 Instant, GPT-5.1 Thinking and GPT-5.1 Pro from ChatGPT on March 11, automatically moving old conversations onto newer models. That makes routing, fallback behavior and retirements part of the commercial product, not just a research update. (help.openai.com) The company is also separating who pays with money from who pays with attention. ChatGPT release notes on April 16 said ads were rolling out to Free and Go users in Australia, New Zealand and Canada, while Plus, Pro, Business, Enterprise and Education remained ad-free. (help.openai.com) A separate ads FAQ says the ad test began in the United States on February 9, 2026, can show a single sponsored unit below a response, and does not run in Temporary Chats, after image generation, or for users OpenAI knows or predicts are under 18. (help.openai.com) For Business customers, the pricing split is even more explicit. OpenAI’s billing guide says a workspace can buy fixed-cost standard ChatGPT seats or usage-based Codex-only seats, with Codex seats carrying no fixed monthly price but requiring workspace credits for activity. (help.openai.com) The result is a ChatGPT lineup where model access, fallback models, coding credits and ads are all being tuned by plan. OpenAI’s latest documentation reads less like a feature list and more like a rate sheet for a metered software platform. (help.openai.com 1) (help.openai.com 2)

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.