Codex 3.0 gets free tier

OpenAI just made its Codex coding agent available on the ChatGPT Free plan — at least for a limited time. That is the real news here, not the YouTube reaction. And it matters because Codex is not just “AI that writes snippets.” It is OpenAI’s agent-style coding product — the one built to take tasks, work in cloud sandboxes, and help with reviews, fixes, and repo-level work. (developers.openai.com) ### What actually changed? The official Codex pricing page now shows Codex as included in ChatGPT Free, Go, Plus, Pro, Business, Edu, and Enterprise plans. The Free tier is framed pretty narrowly — “Explore Codex capabilities on quick coding tasks” — which tells you this is a sampler, not a generous unlimited plan. The Help Center says the Free inclusion is temporary, while paid tiers get broader ongoing access. (developers.openai.com) ### So is this “Codex 3.0”? Probably not in the official product naming. OpenAI’s public Codex materials talk about Codex as the product and describe the underlying models and plans, but the pages surfaced here do not announce something formally branded “Codex 3.0.” What did change recently is packaging and access — especially pricing, plan entitlements, and the spread of Codex acro(developers.openai.com)ollout, but the version label looks unofficial. (developers.openai.com) ### What is Codex, exactly? Codex is OpenAI’s cloud software engineering agent. OpenAI describes it as able to write features, answer questions about a codebase, fix bugs, and propose pull requests, with each task running in its own cloud sandbox loaded with your repository. That is the important distinction — this is closer to delegated coding work than to classic inline autocomple(developers.openai.com)ws. (openai.com) ### Why does free access matter so much? Because agent tools have a higher trial cost than chat tools. You usually need a repo, a use case, and enough allowance to see whether the thing can actually finish work. Putting Codex on Free lowers that friction. A solo developer, student, or curious team member can now test the basic workflow before asking for a paid plan. That changes adoption at the top of the funnel — even if the free limits are tight. (developers.openai.com) ### What do paid plans still get? More room to work, basically. The pricing page says Plus includes Codex across the web, CLI, IDE extension, and iOS, plus cloud integrations like automatic code review and Slack integration. Pro gets much higher usage than Plus, and OpenAI is temporarily honoring elevated 5-hour limits through May 31, 2026. Business and Enterprise also have flexible pricing and even Codex-only seat options now. (developers.openai.com) ### Is there a catch? Yes — limits and entitlements matter more than the headline. OpenAI has been shifting Codex pricing toward token- and credit-based usage instead of simple message counts, and the free plan is explicitly positioned for quick tasks. So “free tier” does not mean full-strength agent access. It means you can get in the door and test the product. Serious daily use still points toward paid plans. (help.openai.com) ### Why are people reacting so fast? Because this is the kind of change that can widen a product’s audience overnight. Codex has been positioned as OpenAI’s answer to real engineering workflows — multi-agent, repo-aware, cloud-executed work. Once that becomes tryable without paying upfront, creators jump on it, developers experiment, and teams start ask(help.openai.com)tion videos. (openai.com) ### Bottom line The important part is simple — OpenAI did not just tweak a model. It opened the door to Codex on ChatGPT Free, even if only in a limited way for now. That makes agent-style coding easier to test, easier to demo, and much easier to spread. (developers.openai.com)