OpenAI gives Codex boosts to 8,000

Coding assistants are starting to look less like a chatbot feature and more like cloud infrastructure. That’s the real story here. OpenAI didn’t just hand out more Codex access after the GPT‑5.5 rollout — it also made Codex easier to reach across ChatGPT plans and clearer to buy more of when you hit limits. (help.openai.com) ### What changed for regular ChatGPT users? Codex is now included by default with ChatGPT Plus, Pro, Business, and Enterprise/Edu plans. For a limited time, OpenAI also added Codex to ChatGPT Free and Go, and it says the other plans get 2x Codex rate limits during the promo window. That is a meaningful change because Codex is not just “write me a function” chat — it spans the Codex app, CLI, IDE extension, and web workflows. (help.openai.com) ### Why do rate limits matter so much? Because Codex is expensive in a different way from normal chat. OpenAI’s own help docs say usage varies with task size, context length, execution mode, and model choice, and its rate card says average spend can land around $100 to $200 per developer per month, with big variance. In other words, one heavy de(help.openai.com)mpts. (help.openai.com) ### So is this still a subscription product? Basically, half yes and half no. OpenAI is keeping the subscription wrapper, but underneath it has built a hybrid system: included limits first, then optional credits after that. Plus and Pro users can buy credits when they hit their Codex allowance, without upgrading plans, and OpenAI describes the w(help.openai.com)help.openai.com) ### Why is that a bigger deal than it sounds? Because it changes the mental model. A normal chat subscription feels like buffet pricing — pay once, use the feature. Codex is moving toward utility pricing — you get a bundle, then you meter the overflow. That makes sense for agentic coding, where compute can spike fast, but it also means developers and managers now have to think about burn rate, not just seat count. (openai.com) ### What about business teams? OpenAI has been quietly building the admin side for this. ChatGPT Business now has workspace analytics that include Codex usage and credits spent across the selected time range. Enterprise and Edu docs also describe separate Codex seats that don’t carry a fixed monthly per-user price and instead use workspace credits, while standard ChatGPT seats keep ba(openai.com)luded limits. (help.openai.com) ### Why bundle Codex into Free and Go at all? Because OpenAI wants more people to try the coding agent before asking them to pay for overflow. That is the same logic behind free cloud credits or starter API quotas — lower the friction, let users build a habit, then monetize the heavier workloads. OpenAI’s own engineering post more or less spells thi(help.openai.com)ng instead of bouncing. (help.openai.com) ### Does model choice change the economics? Yes — a lot. OpenAI’s Codex changelog says GPT‑5.4 mini uses about 30% as much of the included limits as GPT‑5.4 for comparable routine local messages, which can make allowance last roughly 3.3x longer. So “more access” is not just about plan tier. It’s also about steering users toward cheaper models for routine work and reserving the heavy models for harder jobs. (developers.openai.com) ### Bottom line? This looks like a promo on the surface, but the deeper move is commercial. OpenAI is turning Codex into a metered software-development service with subscriptions on top — closer to cloud compute than classic SaaS. If that model sticks, teams won’t just ask whether Codex is good. They’ll ask whether its output is worth the monthly burn. (help.openai.com)sage-in-chatgpt-freegopluspro-sora))