AI platforms move to monetise
OpenAI has added a $100/month Pro tier aimed at heavy Codex (coding) users, a sign vendors are slicing offerings to match distinct commercial needs (techcrunch.com). At the same time, Microsoft appears to be embedding Anthropic and OpenAI models directly inside Microsoft 365 Copilot so enterprises can pick models within the same workflow (windowsreport.com). Separately, OpenAI is reportedly cautious about releasing cybersecurity‑focused models publicly because of the damage potential, which highlights the reputational and distribution constraints around high‑risk AI products (axios.com).
OpenAI just inserted a new $100-a-month step into ChatGPT’s pricing ladder, between the $20 Plus plan and a higher $200 Pro plan that TechCrunch said is still available even though it is no longer shown on the pricing page. The new tier is aimed at people who use Codex, OpenAI’s coding agent inside ChatGPT, heavily enough to outgrow the cheaper plan. (msn.com) That is a pricing change, but it is also a product map. OpenAI is separating casual chat users from developers who want long coding sessions, the same way cloud companies long ago stopped selling one generic server to everyone. (macrumors.com) Microsoft is making the same market more segmented from the other side. In Microsoft 365 Copilot, the company is now showing Anthropic Claude models and OpenAI GPT models inside the same workplace software, so a company can stay in Word, Excel, Outlook, and SharePoint while switching the engine underneath. (windowsreport.com) Microsoft’s own Tech Community post says this multi-model setup is part of Copilot Cowork, which can generate briefing documents, presentations, and spreadsheets from one prompt while pulling from email, calendar, and SharePoint data through a feature called Work IQ. That turns model choice into a menu item inside a work flow, not a separate vendor decision made months earlier by an information technology department. (techcommunity.microsoft.com) Put those two moves together and the shape of the business gets clearer. OpenAI is charging different prices for different intensity of use, while Microsoft is packaging rival models together so the customer pays for convenience and integration instead of loyalty to one lab. (msn.com) (windowsreport.com) There is a second force pushing companies toward narrower products: risk. Axios reported on April 9 that OpenAI is finalizing a cybersecurity product with advanced hacking-related capabilities, but plans to release it only to a small set of partners instead of the general public. (axios.com) That caution did not appear out of nowhere. Axios had already reported in December 2025 that OpenAI expected upcoming frontier models to pose “high” cybersecurity risk as their ability on hacking tests rose quickly, which means the company has been preparing for a world where some models are too dangerous for open retail release. (axios.com) So the artificial intelligence market is starting to look less like one app store and more like an airport. Some products are sold to everyone at a posted price, some are sold to heavy travelers with lounge access, and some never get past the security door unless the operator knows exactly who is walking in. (macrumors.com) (axios.com) The old question was which model was smartest. The new question is which model can be sold safely, bundled cleanly, and priced tightly enough that a freelancer, a software team, and a Fortune 500 company all see a different shelf with a different label on it. (windowsreport.com) (msn.com)
