Home / Codex pricing
Codex pricing and usage limits
Codex, OpenAI’s coding agent, comes with ChatGPT plans. Free and Go users get it in the desktop app with GPT-6 Luna, as it rolls out; the full version starts with Plus at $20 a month, with Pro from $100 and Business at $20 a user. You can also sign in with an API key and pay per token.
Checked October 20, 2026 against OpenAI’s Codex pricing and sign-in pages · API prices checked October 13, 2026
| Plan | Price | Codex on the plan |
|---|---|---|
| Free | $0 | GPT-6 Luna at Standard speed in the ChatGPT desktop app, as it rolls out |
| Go | $8 | The same as Free |
| Plus | $20 | Codex on the web, in the CLI, in the IDE extension and on iOS; automatic code review and Slack; GPT-6.1 Sol and GPT-6 Luna; more usage with ChatGPT credits |
| Pro | From $100 ($100, $200, $500) | Everything in Plus, with no five-hour limit at present; Ultrafast mode on Pro $500 |
| Business | $20 a user, billed yearly ($25 monthly) | For 2 or more users: admin controls, single sign-on, larger cloud machines, no training on business data by default, and workspace credits |
| Enterprise & Edu | Contact sales | Priority processing, enterprise security, audit logs, and data retention and residency controls |
| API key | Per token | The CLI, SDK and IDE extension at standard API prices; no cloud features |
ChatGPT Work and Codex share the same usage, credits and limits, OpenAI says.
ChatGPT plans that include Codex, by price
- FreeFree
- Go$8 a month
- Plus$20 a month
- Business, per user$25 a month$20 billed yearly
- Profrom $100 a month
US prices a month. Enterprise and Edu are priced by OpenAI’s sales team.
Codex usage limits
Plans don’t charge per token. Instead they include usage measured in five-hour periods, and weekly limits may also apply. OpenAI’s estimates of local messages per five hours on Plus and on a standard Business seat:
| Model | Local messages per 5 hours |
|---|---|
| GPT-6 Astra | 5 to 45 |
| GPT-6.1 Sol | 15 to 160 |
| GPT-6 Sol | 15 to 150 |
| GPT-6 Luna | 350 to 3,000 |
The range is wide because a small fix uses a fraction of what a long session over a large project does: the model, the context, reasoning, tools and caching all change how fast you use your allowance, so prompt length alone isn’t a good guide. Cloud tasks may use more than local messages. Pro plans have no five-hour limit at present. Enterprise and Edu plans without flexible pricing get the same per-seat limits as Plus for most features.
Faster modes use your allowance faster. Fast mode counts 2.5 times against included usage and 2 times against purchased credits. Ultrafast, on Pro $500 and eligible Enterprise and Edu plans, counts 8 times and 6 times. Generating images uses limits about 3 to 5 times faster.
Your current limits and reset times are on the Codex usage dashboard; in the CLI, /status shows what’s left.
When you run out: credits
Plus and Pro users who reach their limit can buy ChatGPT credits and keep working, and Business, Enterprise and Edu workspaces with flexible pricing can buy workspace credits. Each model uses credits at these rates, per million tokens at Standard speed:
| Model | Input | Cached input | Output | API price, in / out |
|---|---|---|---|---|
| GPT-6 Astra | 250 | 25 | 1,250 | $10.00 / $50.00 |
| GPT-6.1 Sol | 50 | 2.5 | 250 | $2.00 / $10.00 |
| GPT-6 Sol | 50 | 5 | 250 | $2.00 / $10.00 |
| GPT-6 Luna | 2.5 | 0.25 | 12.5 | $0.10 / $0.50 |
OpenAI doesn’t give one dollar price for credits on its Codex page: what credits cost depends on your plan or agreement. Every credit rate in the table is 25 times the model’s API price in dollars, so you can compare a credit offer with paying per token: credits match API prices when 25 credits cost $1. Codex credit billing has no separate charge for writing to the cache.
Codex with an API key
Instead of signing in with ChatGPT, you can sign in to the ChatGPT desktop app, the Codex CLI or the IDE extension with an API key from the OpenAI dashboard. Codex then bills your OpenAI Platform account at standard API prices instead of using a plan’s included usage. OpenAI suggests this for automated jobs such as CI pipelines. Codex cloud and the cloud features, like GitHub code review and Slack, need a ChatGPT sign-in.
For complex coding, OpenAI’s Codex docs recommend GPT-6.1 Sol, which OpenAI describes as close to Astra’s performance at a lower cost, and GPT-6 Luna for focused, repeatable tasks. Their API prices, per million tokens:
| Model | Input | Cached input | Output |
|---|---|---|---|
| GPT-6 Astra | $10.00 | $1.00 | $50.00 |
| GPT-6.1 Sol | $2.00 | $0.10 | $10.00 |
| GPT-6 Luna | $0.10 | $0.01 | $0.50 |
From OpenAI’s pricing page, checked October 13, 2026. Prompts over 272,000 tokens cost more; see OpenAI API pricing.
Estimate pay-as-you-go cost
Choose models
- Turn on JavaScript to see the costs. The price tables below work without it.
A plan or an API key?
A plan fixes what you pay each month and includes the cloud features, but its limits are estimates that vary from task to task. An API key charges for exactly the tokens you use, with no five-hour window (the API’s own rate limits still apply), and suits scripts and automated jobs. If you code with Codex most days, start with Plus and watch the usage dashboard; if you hit the limit often, compare Pro with buying credits or using an API key for the extra work.
Anthropic’s Claude Code works the same two ways, with paid Claude plans or per-token billing: see Claude Code pricing. Through the APIs, GPT-6.1 Sol has the same list price as Claude Sonnet 5.5: $2.00 in and $10.00 out per million tokens.
Questions
How much does Codex cost?
It comes with ChatGPT: $20 a month on Plus, from $100 on Pro, and $20 a user on Business when billed yearly. With an API key, you pay OpenAI’s per-token prices instead.
Is Codex free?
There is a limited version on the Free and Go plans: GPT-6 Luna at Standard speed in the ChatGPT desktop app, as OpenAI rolls it out. Image generation in Codex isn’t available on Free.
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.
When do Codex limits reset?
Included usage works in five-hour periods, and weekly limits may also apply. The usage dashboard shows your reset times.
What happens when I reach the limit?
Codex finishes the turn it’s working on, subject to fair use. Then you can buy credits (Plus and Pro), use workspace credits (Business, Enterprise and Edu with flexible pricing), wait for the reset, or run extra local chats with an API key at API prices.
Which model should I use in Codex?
OpenAI recommends GPT-6.1 Sol for complex coding and agent work where your plan has it, and GPT-6 Luna for focused, repeatable tasks. GPT-6.1 Sol wasn’t included on the Free and Go plans at launch. GPT-5.5 left Codex for ChatGPT sign-ins on October 14, 2026; with an API key it’s still available.
Sources
- OpenAI: Codex pricing (plans, usage limits, credit rates, speed multipliers), read October 20, 2026
- OpenAI: Codex authentication (signing in with an API key), Codex models and speed modes, read October 20, 2026
- OpenAI: API pricing and Anthropic: pricing (token prices, checked October 13, 2026)