AI coding tools · 2026
Codex pricing, plan by plan
OpenAI’s Codex isn’t sold on its own: it comes with every ChatGPT plan, or you can pay per token with an API key. Here’s what each plan includes, how far it goes, and what the API route costs.
Plans from developers.openai.com/codex/pricing, checked November 4, 2026. Model prices from OpenRouter, November 8, 2026
| Plan | Price | Where Codex runs | Models and usage |
|---|---|---|---|
| Free | $0 | The desktop app, subject to rollout | GPT-6 Luna at Standard speed |
| Go | $8 a month | The desktop app, subject to rollout | GPT-6 Luna at Standard speed |
| Plus | $20 a month | Web, CLI, IDE extension and iOS, plus cloud integrations like automatic code review and Slack | GPT-6.1 Sol and GPT-6 Luna. Buy credits if you run out |
| Pro | $100, $200 or $500 a month | Everything in Plus, with no five-hour limit for now. Ultrafast mode on Pro $500 | As Plus, with more usage at each price |
| Business | $20 per user a month billed yearly, $25 monthly (2+ users) | Larger cloud machines, SAML SSO, MFA and admin controls | Same limits per seat as Plus. Buy shared workspace credits on flexible pricing |
| Enterprise and Edu | Contact sales | Priority processing, SCIM, EKM, RBAC, audit logs, data residency | Usage scales with credits on flexible pricing |
| API key | Per token | CLI, SDK and IDE extension only. No cloud features like GitHub code review or Slack | Whatever your API key can use |
ChatGPT Work and Codex share the same usage. OpenAI’s pricing page doesn’t say how much more usage each Pro tier adds.
How far a plan goes
A message isn’t a fixed amount of work, so OpenAI gives ranges.
| Model | Local messages per 5 hours |
|---|---|
| GPT-6 Astra | 5–45 |
| GPT-6.1 Sol | 15–160 |
| GPT-6.1 Terra | 35–285 |
| GPT-6 Sol | 15–150 |
| GPT-6 Luna | 350–3,000 |
OpenAI’s estimates for Plus and Standard Business, not fixed limits. Weekly limits may also apply, and cloud tasks may use more of your allowance. Pro currently has no five-hour limit. Fast mode uses limits 2.5× as fast, and Ultrafast 8×.
The range is the point. One GPT-6.1 Sol message can use ten times what another does. Model, context, reasoning, tools and caching all count. To stretch a plan: use GPT-6 Luna for small tasks, keep AGENTS.md short and turn off MCP servers you aren’t using, since each one adds context to every message.
On an API key: what an hour costs
| Model | Input /1M | Output /1M | Per hour | 1 h a day | 4 h a day |
|---|---|---|---|---|---|
| GPT-6 Astra | $10 | $50 | $9.45 | $198 | $794 |
| GPT-6.1 Sol | $2 | $10 | $1.62 | $34.02 | $136 |
| GPT-6.1 Terra | $1 | $5 | $0.81 | $17.01 | $68.04 |
| GPT-6 Sol | $2 | $10 | $1.89 | $39.69 | $159 |
| GPT-6 Luna | $0.10 | $0.50 | $0.09 | $1.98 | $7.94 |
21 working days. An hour of agent coding is about 3M input tokens (90% of them cache reads) and 60K output tokens; your use can be several times higher or lower. Use your own numbers. OpenAI says API prices don’t tell you what a plan includes, so don’t use this table to guess a subscription’s limits.
Credits per million tokens
Past your limits on Plus, Pro, or a credit-based Business plan, usage draws on credits at these rates.
| Model | Input | Cached input | Output |
|---|---|---|---|
| GPT-6 Astra | 250 | 25 | 1,250 |
| GPT-6.1 Sol | 50 | 2.5 | 250 |
| GPT-6.1 Terra | 25 | 1.25 | 125 |
| GPT-6 Sol | 50 | 5 | 250 |
| GPT-6 Luna | 2.5 | 0.25 | 12.5 |
Standard speed. There’s no separate cache-write charge in credits. On purchased credits, Fast mode costs 2× these rates and Ultrafast 6×.
Codex, Claude Code and Cursor at the same price
| Monthly | Codex (ChatGPT) | Claude Code | Cursor |
|---|---|---|---|
| $0 | Free: GPT-6 Luna in the desktop app, subject to rollout | Not on the free plan | Free: limited agent and Chat requests |
| $8 | Go: GPT-6 Luna in the desktop app, subject to rollout | – | Plus: Cursor’s own models only |
| $20 | Plus: web, CLI, IDE and iOS, GPT-6.1 Sol | Pro: Sonnet and Opus, within limits | – |
| $30 | – | – | Premium: 4× Plus; Claude and GPT at list price on top |
| $100 | Pro $100: no five-hour limit for now | Max: 5× Pro, plus $100 a month in API credits | Super: 15× Plus; Claude and GPT on top |
| $200 | Pro $200: no five-hour limit for now | Ultra: 10× Pro a week, no 5-hour limit, plus $200 a month in API credits | Ultra: 50× Plus; Claude and GPT on top |
| $500 | Pro $500: adds Ultrafast mode | – | – |
Checked November 4, 2026. The “×” figures compare within one vendor only. Cursor vs Claude Code in detail.
Related
Codex pricing questions
How much does Codex cost?
Codex comes with ChatGPT: Free ($0), Go ($8 a month), Plus ($20), Pro ($100, $200 or $500) and Business ($20 per user a month billed yearly, $25 monthly). Enterprise and Edu are priced by OpenAI’s sales team. You can also skip the plans and pay per token with an API key.
Is Codex free?
There is a free way in: the ChatGPT Free plan includes GPT-6 Luna at Standard speed in the desktop app, subject to rollout. For Codex on the web, in the CLI and in your IDE, you need Plus or above, or an API key.
Is Codex included with ChatGPT Plus?
Yes. Plus includes Codex on the web, in the CLI, in the IDE extension and on iOS, with GPT-6.1 Sol and GPT-6 Luna. OpenAI estimates 15–160 local messages per five hours on GPT-6.1 Sol. Weekly limits may also apply.
How much does the Codex CLI cost?
The CLI comes with Plus and above. Or sign in with an API key and pay per token at API prices, which suits automation in shared environments like CI. Cloud features like GitHub code review and Slack don’t work on an API key.
What happens when I hit the Codex limit?
A turn already running can finish, subject to fair use. After that, Plus and Pro users can buy credits to keep going, or anyone can run extra local chats with an API key at API rates. Business, Edu and Enterprise plans on flexible pricing can buy workspace credits.
What is a Codex credit?
Credits pay for usage past your included limits on credit-based plans. Each model uses a set number of credits per million tokens (the table above). What a credit costs depends on your plan or agreement. OpenAI says a typical GPT-5.6 Sol task may use 2–15 credits.
Is Codex cheaper than Claude Code?
The plans start at the same price: $20, $100 and $200 on both. Neither gives limits in tokens, so the honest way to compare is to run a week of your own work on each. On an API key, GPT-6.1 Sol lists at $2 in and $10 out per 1M tokens, and Claude Sonnet 5.5 at $2 and $10. Claude Code pricing.
How do I see what AI coding costs per product?
Subscriptions show up in Glow from your Ramp card, and API spend by model and key from OpenRouter, next to each product’s revenue. Glow doesn’t read OpenAI’s own usage API yet. Track AI spend.
Know what your AI stack costs.
Glow shows each product’s AI tools and API bill against its revenue, so you know which app is worth the tokens.