Scope: two OpenAI documentation pages, "Pricing" and "Speed," checked October 6, 2026 (20:05 KST). Your plan, workspace, client and account balance are UNKNOWN to us. We have not bought credits, changed a speed setting, benchmarked Fast mode or seen anyone’s bill for this article.
Short answer: Before you budget cash for Codex Fast mode, find out which meter your session is drawing from. OpenAI lists two different Fast factors, both relative to Standard mode for the same supported model: Fast uses included subscription limits "at 2.5x the Standard rate," while "Purchased credits and Enterprise pay-as-you-go usage are billed at 2x the Standard rate." With an API key, neither factor applies. And OpenAI is explicit that "These billing multipliers don’t describe speed increases."
Which balance is being used?
Fill in the middle column from your own account. Leave a cell UNKNOWN until you have checked it, and never enter 0 for a value you haven’t seen. Don’t write account emails, card numbers or API keys here.
| Your meter | What to record from your account | What not to infer |
|---|---|---|
| Included subscription allowance | Your plan, the same model and speed mode, your actual remaining allowance and reset time: UNKNOWN | A dollar price, or a fixed number of tasks, from a usage percentage |
| Purchased credits / applicable Enterprise pay-as-you-go | Your plan’s or agreement’s credit terms, the same model and speed mode, your actual credit balance: UNKNOWN | One universal dollar value per credit |
| API key | The API rates that apply to your key and your own billing records: UNKNOWN | ChatGPT credit multipliers, or included-task estimates |
The three rows don’t convert into each other. OpenAI says "Credit purchase prices and applicable discounts depend on your plan or agreement," so this guide gives no dollar figure per credit. For API keys, OpenAI says Codex "uses API token pricing instead, and ChatGPT credit multipliers don’t apply." It also warns that "API token prices are separate from subscription usage; don’t use them to estimate included tasks."
Read the Fast factor correctly
| Meter | Fast relative to Standard (same supported model) | Evidence |
|---|---|---|
| Included subscription usage | 2.5x | OpenAI Pricing and Speed pages, checked 2026-10-06 |
| Purchased credits / Enterprise pay-as-you-go usage | 2x | OpenAI Pricing and Speed pages, checked 2026-10-06 |
| API key | Not applicable: API token pricing applies | OpenAI Speed page, "API billing" |
These factors are about billing, not time. A 2.5x factor doesn’t mean a task finishes 2.5x faster, and it doesn’t tell you how many tasks your allowance covers. OpenAI adds that "Credit rates alone don’t determine how quickly you use included subscription limits." For Enterprise, OpenAI says that if "your Enterprise agreement specifies usage-based billing in USD," you use the Enterprise USD rate card "and your agreement" rather than the credit-based rate card.
A before-pay checklist
- Sign-in and billing route. Are you signed in with ChatGPT or using an API key? OpenAI lists Fast mode in "the ChatGPT desktop app, Codex CLI, and IDE extension when you sign in with ChatGPT."
- Plan, model and client. Fast mode covers supported models only, "where available," and "Access depends on your plan, client, workspace settings, and rollout." Record whether your own model and client show it. If not, write UNKNOWN.
- Actual balance and reset. Open your usage dashboard and write down your current limits and reset time. Record any purchased-credit balance from wherever your plan or workspace shows it. If you can’t confirm it, write UNKNOWN.
- Does waiting matter? Ask whether latency actually changes the outcome of the work. If it doesn’t, Standard mode is one editorial option to consider. That’s a suggestion, not a measured result, and we haven’t changed any setting.
FAQ
Does 2.5x usage mean 2.5x faster completion? No. OpenAI says the billing multipliers "don’t describe speed increases." The exact speed gain for your model and task is UNKNOWN here.
Does an API key use the same credit multiplier? No. With an API key, Codex uses API token pricing, and ChatGPT credit multipliers don’t apply.
Can API token prices estimate the tasks in my subscription? No. OpenAI says not to use them to estimate included tasks. Usage also varies by task: "Model choice, context, reasoning, tool use, retrieval, and caching all affect usage."
Where do I check actual balance and resets? In the usage dashboard. OpenAI says "You can find your current limits in the usage dashboard," and in an active Codex CLI session "you can use /status." The same page says to "check your usage dashboard for current limits and reset times."
Next step
Before you buy credits or upgrade, check your own usage and billing route in the Codex usage dashboard. Factors checked October 6, 2026 against OpenAI’s Pricing and Speed pages. OpenAI can change these pages, so re-check them before you rely on this guide.