OpenAI rollout remains staged
OpenAI is rolling enterprise features out stepwise rather than universally: Custom Actions now work with GPT-4o and 4.1 on the web for Plus, Pro and Team users, while Enterprise and Education access is still labelled ‘coming soon’. At the same time, OpenAI published a Codex rate card detailing credits and plan pricing, underlining that enterprise adoption is moving into procurement-driven decisions. (help.openai.com) (help.openai.com).
OpenAI is not flipping one giant switch for business customers. It is adding features the way a landlord opens floors in a new building: one group gets keys now, another group gets a “coming soon” sign. (help.openai.com) The latest example is Custom Actions inside ChatGPT on the web. OpenAI’s release notes say they now work with GPT-4o and GPT-4.1 for Plus, Pro, and Team users, while Enterprise and Education access is still listed as coming soon. (help.openai.com) Custom Actions are the part that lets ChatGPT do something outside the chat window, like calling another app or service instead of only writing text. They turn the model from a typing assistant into a worker that can actually trigger steps in a workflow. (help.openai.com) That split matters because Team, now renamed ChatGPT Business, is OpenAI’s self-serve workplace tier, while Enterprise and Education are the contract-heavy tiers with admin controls and procurement reviews. OpenAI’s own billing pages now separate those groups much more clearly than before. (help.openai.com 1) (help.openai.com 2) The pricing side moved at the same time. On April 2, 2026, OpenAI changed Codex pricing for new and existing ChatGPT Business customers and for new Enterprise customers from per-message estimates to token-based pricing, which is the same basic unit used in application programming interface billing. (help.openai.com) A token is a small chunk of text, like billing by ingredients instead of by plate. OpenAI’s new rate card lists separate charges per 1 million input tokens, cached input tokens, and output tokens, with GPT-5.4 priced at 62.50 credits for input and 375 credits for output. (help.openai.com) OpenAI also says Fast mode costs 2 times as many credits, and code review uses GPT-5.3-Codex. The same rate card says average Codex spend runs about $100 to $200 per developer per month, with big swings depending on model choice, automations, and how many instances are running. (help.openai.com) The company is also changing how seats work. As of April 2, 2026, ChatGPT Business and ChatGPT Enterprise can use two seat types: a standard ChatGPT seat and a Codex-only seat, while Education does not get those seat options. (help.openai.com) On the self-serve Business plan, a standard seat now costs $25 a month on monthly billing or $20 a month on annual billing, and Codex-only seats have no fixed monthly price but require workspace credits. OpenAI also says self-serve Business still does not support invoices, purchase orders, bank transfers, or net terms. (help.openai.com) That is the shape of the rollout. Consumers and small teams get more capability first, while larger organizations get more pricing knobs, spend controls, and contract pathways before they get every new feature at the same time. (help.openai.com 1) (help.openai.com 2) So the story is not just that OpenAI shipped another feature. It is that ChatGPT is being split into two motions at once: faster product rollout for lighter plans, and slower, meter-based rollout for the customers who need approvals, budgets, and someone in procurement to sign the paper. (help.openai.com 1) (help.openai.com 2)