OpenAI Tightens Pricing and Ads
OpenAI is updating its Codex pricing and signalling that Enterprise and other plans will migrate to new rates in the coming weeks, a sign the company is making pricing for developer tools more conventional (help.openai.com). At the same time ChatGPT will begin testing ads in the U.S. on free tiers while keeping Pro and paid plans ad-free, and the service recently reset Codex usage limits after reaching heavy weekly usage, highlighting both monetisation and capacity pressure (help.openai.com) (technobezz.com).
OpenAI is changing two parts of the same machine at once: the coding tool is moving toward pay-as-you-go billing, and the chat app is starting to sell ad space on its cheapest tiers in the United States. The company’s own help pages say Codex pricing changed on April 2, 2026, and ads began testing in the United States on February 9, 2026. (help.openai.com 1) (help.openai.com 2) Codex is OpenAI’s coding assistant, and until now a lot of its pricing worked like a meal plan: you got rough message allowances instead of a meter reading for every bite. OpenAI says new and existing ChatGPT Business customers and new ChatGPT Enterprise customers are now billed by tokens, which are the chunks of text the model reads and writes. (help.openai.com) That shift matters because token pricing is the standard language of the wider artificial intelligence developer market. OpenAI says the new card replaces average per-message estimates with credits per 1 million input tokens, cached input tokens, and output tokens, so teams can see more directly what a long prompt, a reused context window, or a big code generation actually costs. (help.openai.com) The new table is blunt about where the meter runs fastest. On the April 2026 rate card, GPT-5.4 costs 62.50 credits per 1 million input tokens and 375 credits per 1 million output tokens, and OpenAI says Fast mode burns credits at 2 times the normal rate. (help.openai.com) OpenAI is not flipping every customer at once. Its help center says existing Plus, Pro, and Enterprise or Education customers are still on the legacy rate card for now, and those plans will move to the new rates “in the upcoming weeks.” (help.openai.com) The company is also nudging teams toward a more ordinary software budget. The developer pricing page says Business customers can assign usage-based Codex seats, and the Business release notes say there is no minimum number of Codex seats because usage requires workspace credits instead of a fixed bundle alone. (developers.openai.com) (help.openai.com) At the same time, the consumer side is getting a different revenue lever. OpenAI’s ads policy says ads may appear for logged-in adults on the Free and Go plans in the United States, while Plus, Pro, Business, Enterprise, and Education accounts stay ad-free. (help.openai.com 1) (help.openai.com 2) OpenAI is drawing a hard wall between ads and answers because that is the part users will distrust first. The company says ads run on separate systems from the chat model, advertisers cannot shape responses, and accounts identified as under 18 are excluded from the ad test. (help.openai.com) Then came the capacity signal. OpenAI’s Codex rate card says average usage runs about $100 to $200 per developer per month with wide variation, and a report published on April 9 said the company reset Codex limits after the tool reached 3 million weekly users. (help.openai.com) (technobezz.com) Put together, the pattern is simple: expensive coding work is being moved onto a utility bill, and mass-market chat is being asked to subsidize itself with ads unless you pay to leave the billboard behind. That is a more conventional internet business than the one OpenAI sold at launch, but it fits a product now serving hundreds of millions of users and increasingly heavy coding workloads. (help.openai.com 1) (help.openai.com 2)