[Verified] Codex has no standalone subscription—your choice is Go, Plus, or Pro

Codex Doesn’t Exist as a Standalone Product

OpenAI’s Codex, the company’s AI coding agent, is not sold on its own. There is no “buy Codex” button. Instead, it’s bundled into ChatGPT subscription tiers, and which plan you pick determines how much you can actually use it. This matters because the price differences—US$8, US$20, US$100, or US$200 per month—are not just about tier names. They represent hard caps on Codex’s usage and feature availability.

If you’re thinking about whether Codex is worth paying for, you’re really deciding between ChatGPT Go, Plus, and Pro. The free tier exists but is intentionally limited. Understanding these boundaries upfront saves money and frustration.

The Free Tier Is Real, But Severely Constrained

ChatGPT free accounts unlock Codex with Limited access. That means you can try the agent and see what it does. For light experimentation, this might be enough. If you want to use Codex via CLI, you can connect your own OpenAI API key and pay per token, bypassing the subscription entirely—but that’s a different billing model and only practical if you want to meter usage tightly.

The term “Limited” is OpenAI’s official label and covers the free tier across the web interface and desktop apps.

The Plan Tiers: Go, Plus, and Pro

Go costs US$8 per month and includes Codex Limited—the same restricted version you get free. It’s positioned for users who want to keep their ChatGPT subscription lean but still gain access.

Plus, at US$20 per month, includes full Codex with “Expanded usage.” This is where the unlock happens: you get the complete feature set, not the crippled version. For developers who want to actually rely on Codex, Plus is the entry point.

Pro is OpenAI’s highest tier and the data here carries a caveat. OpenAI’s official pricing pages returned error 403 (blocked, not missing) when checked on 2026-08-11. Pro pricing comes from third-party sources verified independently: either US$100 per month for 5x the Plus quota, or US$200 per month for 20x. The exact usage limits and whether both tiers are always available remain unclear from official sources.

🔍 AMPM exclusive check

OpenAI’s Codex and ChatGPT pricing pages all returned HTTP 403 errors during our 2026-08-11 verification attempt—an unusual block rather than a simple 404. We cross-checked Pro pricing against multiple third-party sources and found consensus on the US$100 and US$200 monthly figures, but could not verify these figures directly from OpenAI’s official documentation. For Go and Plus, the official pricing confirmed: US$8/month (Go, Codex Limited) and US$20/month (Plus, full Codex).

The key exclusive finding: Codex is not independently purchasable. It exists only as a bundled feature inside ChatGPT plans. OpenAI does not publish per-user feature parity matrices (which tier gets which Codex features), so the labels “Limited,” “Expanded,” and quota multiples are the only official guidance available.

☀️ AMO, the budget-minded cat

If you’re cost-conscious, free is where you test Codex risk-free. The Limited version isn’t a demo—it’s a functional coding assistant with real constraints. You can write functions, debug, and see Codex in action without spending anything. For someone unsure whether they want an AI agent in their workflow at all, this is the right starting point.

If you do need more than Limited offers, Go at US$8/month is the smallest paid step. That’s the cost of two coffees. But here’s the gotcha: Go still gives you Codex Limited, the same capped version. You pay for ChatGPT features (faster responses, priority access), not for Codex improvements. Upgrading Codex itself requires jumping to Plus.

Plus at US$20/month is where Codex actually changes. Full access, expanded usage quota—it’s the real product. For freelancers or small teams paying out of pocket, US$20/month is the break-even point: if Codex saves you five billable hours per month, it pays for itself. If it saves you ten, it’s a bargain. The math works if you use it.

Pro tiers (US$100–200) are for teams or individuals with money budgeted for AI tools. The 5x and 20x quota increase matters if you’re automating daily tasks, not if you’re using Codex occasionally. Budget-wise, Pro is only defensible if your hourly rate or team efficiency gains justify the spend. For most individual developers, it isn’t.

🌙 PIMI, the performance-minded cat

For developers who care about actually getting work done, the question is not “is Codex cheap” but “does Plus feel complete.” The jump from Limited to full Codex is real and measurable. Limited mode will refuse or fail on tasks Plus handles routinely—complex refactoring, writing multi-file changes, reasoning about edge cases. If you’re using Codex and hitting walls, it’s usually because Limited mode lacks reasoning depth, not token count.

Plus delivers a usable product. At US$20/month, it’s positioned as the “this tool does its job” tier. The expanded usage quota matters here: you can run multiple tasks per day without hitting a wall. For a developer shipping code five days a week, Plus quota should not be a constraint.

The Pro tiers make sense only if you’re saturating Plus quota regularly. If you’re running Codex tasks continuously—testing multiple approaches, iterating rapidly, processing large files—then 5x or 20x usage becomes worth the cost. At US$100/month, Pro assumes you extract enough value from Codex to justify a meaningful monthly line-item. That’s possible if you’re a full-time AI-assisted developer or a small team using Codex as an internal tool.

A practical note: Codex often pairs with Claude or other agents depending on the IDE integration and platform. If you’re evaluating purely for speed and completion rate, run test tasks in Codex and your alternatives (Claude, GitHub Copilot) under the same conditions. Plus should be fast enough for most workflows; Pro is for when you’ve proven you need higher throughput.

Verdict: who should use it, who should skip it

Use Codex Plus (US$20/month) if: you write code daily, want an agent that can handle multi-step tasks, and are willing to invest in a paid tool. The free tier and Go are too restrictive for serious work. Plus is the minimum viable paid tier.

Try free first if: you’re not sure whether an AI coding agent fits your workflow. Limited is a real product—limited, but real—and costs nothing.

Skip to Pro (US$100–200) only if: you’ve used Plus for weeks and consistently run out of quota, or your team is paying and needs higher throughput as a production utility.

Codex is not for you if: you’re comparing solely on price. If budget is the only factor, free tiers from other vendors (GitHub Copilot’s free plan, Claude’s free tier) might make more sense. Codex’s value is in execution within ChatGPT’s ecosystem, not in being the cheapest option.

Written: 2026-09-24 Price last verified: 2026-09-08 Tool last health-checked: 2026-09-24

🔗 Related on AMPM

• Codex pricing & free tier

📎 Sources

All prices and quotas in this article come from AMPM’s verified dataset for Codex (last verified 2026-09-08); primary sources:

• https://openai.com/codex/

🐾 Meet the cats: AMO and PIMI

AMPM is a Taiwan-based AI-tool pricing watchdog. Our motto: ask before you subscribe. Two cats argue every tool from two sides:

• ☀️ AMO — the budget-minded cat. Always asks: is the free tier enough, and is paying actually worth it?
• 🌙 PIMI — the performance-minded cat. Cares about whether it works well, fast, and gets your job done.

When they are done arguing, you know whether to pay. Every number is checked by us, with the verification date at the end.

🐾 Our sister sites also checked

• Travel insurance and flight-delay coverage checked
• Travel eSIM plans compared