You buy OpenAI API credits through your OpenAI account dashboard, not from a store or reseller

OpenAI does not sell "credits" in the traditional sense. Instead, you set up a billing account, add a payment method, and pay for API usage as you go. You are charged based on the tokens you actually use — the small chunks of text that the API processes. The cost varies by which model you use (GPT-4 is more expensive than GPT-3.5, for example) and how many tokens your requests consume.

You can also set spending limits to control your costs, but there is no way to pre-purchase a fixed amount of credits that sit in an account waiting to be used. You pay for what you use, month to month.

Key Takeaways

  • OpenAI charges for API usage based on tokens consumed, not pre-purchased credits, and you set this up in your account dashboard under Billing.
  • Different models cost different amounts per token — GPT-4 Turbo costs more than GPT-3.5-turbo, and pricing changes when OpenAI releases new models.
  • You can set a monthly spending limit in your billing settings to prevent unexpected charges, though this does not may provide protection if you hit the limit mid-request.
  • Your first $5 in usage is often free as a trial period, but this expires after three months, so set up paid billing before then if you plan to keep using the API.
  • You need a valid credit card or debit card to set up billing; some countries and payment methods are not supported.

Setting up billing in your OpenAI account

Log in to your OpenAI account at platform.openai.com. Click your account icon in the top right corner and select "Billing overview." From there, click "Set up paid account" or "Add to paid account" if you already have a free trial running.

You will be asked to enter your billing address and a valid payment method — a credit card or debit card. OpenAI accepts Visa, Mastercard, American Express, and Discover in most countries, but support varies by region. If your country or card type is not supported, you may not be able to set up billing directly.

Once your payment method is saved, you can start making API requests and will be charged based on usage. Check your billing settings regularly to see your current month's spending and any usage breakdown by model.

Understanding token costs and pricing tiers

OpenAI charges per token, not per request. A token is roughly four characters of text — so the word "hello" is one token, and a typical sentence might be 15 to 20 tokens. When you send a request to the API, you are charged for both the tokens you send (the "prompt") and the tokens the model returns (the "completion").

Pricing varies by model. As of early 2024, GPT-3.5-turbo is the cheapest option, costing roughly $0.0005 per 1,000 input tokens and $0.0015 per 1,000 output tokens. GPT-4 Turbo costs significantly more — around $0.01 per 1,000 input tokens and $0.03 per 1,000 output tokens. OpenAI also offers GPT-4o (a faster, cheaper variant) and older models at lower prices.

Pricing changes when OpenAI releases new models or updates existing ones, so check the pricing page at openai.com/pricing before building something you plan to scale. A small test might cost cents, but if you scale to thousands of requests per day, costs can reach hundreds or thousands of dollars monthly.

Setting spending limits and monitoring usage

In your Billing settings, you can set a "Hard limit" — a monthly spending cap that stops new API requests once you reach it. This is not a may provide: if a request is already processing when you hit the limit, you may be charged for it. But it prevents runaway costs from a bug or unexpected traffic spike.

You can also set a "Soft limit" — a warning threshold that sends you an email when you reach it. This gives you time to investigate if your usage is higher than expected.

Check your usage dashboard regularly, especially when you first launch. Look at the "Usage" tab to see which models you are calling most often and which are costing the most. If you are using GPT-4 for tasks that GPT-3.5 could handle, switching models can cut your bill significantly.

Free trial credits and when they expire

New OpenAI accounts typically receive $5 in free trial credits that you can use within three months. These credits cover API usage only — they do not work for ChatGPT Plus or other paid products. Once the three months are up, the credits expire whether you used them or not.

If you want to keep using the API after your trial ends, you must set up paid billing before the credits expire. If you do not, your API requests will fail with a billing error. There is no grace period or warning after the three months — the cutoff is automatic.

Invoicing and payment methods for teams

If you are using the API for a business or team, you can request monthly invoicing instead of automatic card charges. This requires a higher spending threshold — typically a minimum monthly commitment — and you will need to contact OpenAI's sales team to set it up.

You can also add team members to your OpenAI account so they can use the same API key and billing account. Each team member can view usage and spending, but only the account owner can change billing settings or payment methods. If you need separate billing for different projects or departments, you will need separate OpenAI accounts.

What to do if your payment fails or billing is suspended

If your card is declined or your payment fails, OpenAI will attempt to charge it again over several days. During this time, your API requests will fail. Check your email for a payment failure notice — it will tell you which card was declined and ask you to update your payment method.

Log back into your Billing settings and click "Update payment method" to add a different card. Once the new card is saved, your API access should resume within a few minutes. If you see a message that your account is suspended, contact OpenAI support — this usually means there is a billing dispute or a violation of their usage policy, not just a failed payment.

Frequently Asked Questions

Can I buy OpenAI API credits from a third party or reseller?

No. OpenAI only sells API access through your own account at platform.openai.com. Any third party claiming to sell OpenAI credits is either running a scam or reselling access in a way that violates OpenAI's terms. Buy directly from OpenAI.

What happens if I go over my spending limit?

If you set a hard limit, new API requests will be rejected once you reach it, and you will see a billing error. Requests that are already processing may still be charged. If you did not set a limit, you will simply be charged for all usage at the end of the month, with no cap.

Can I get a refund for API usage I did not intend to pay for?

OpenAI rarely refunds API charges, but you can contact support if you believe there was an error — for example, if a bug caused thousands of unintended requests. Provide details of what happened and they will review it. Do not expect a refund for intentional usage, even if you did not realize the cost.

Do I need a business account or tax ID to use the API?

No. You can use the API as an individual with just a valid payment method and billing address. If you are using it for a business, you may want to set up invoicing through OpenAI's sales team, but this is optional and requires higher spending.

What countries can use OpenAI API billing?

OpenAI supports billing in most countries, but some regions and payment methods are restricted. If you cannot add a payment method in your country, check OpenAI's documentation or contact their support team. Some users in restricted regions use a VPN or third-party payment service, but this may violate OpenAI's terms.