GPT-5 Pricing Explained

GPT-5 API pricing is based primarily on the number and type of tokens processed. This is different from ChatGPT subscriptions, where users generally pay for a product plan rather than directly calculating the cost of every prompt.

How the estimate is calculated

For the original gpt-5 API model, OpenAI currently lists standard pricing of $1.25 per million input tokens, $0.125 per million cached input tokens, and $10 per million output tokens.

Suppose your application processes:

What can change the total

• 10 million input tokens • 2 million output tokens

Ignoring caching and other optional charges, the approximate GPT-5 cost would be:

Plan with a realistic usage example

Input: 10 × $1.25 = $12.50

Output: 2 × $10 = $20

Practical checks before you proceed

Total: $32.50

This illustrates an important point: output tokens can contribute significantly more to your bill than input tokens.

Practical checks before you proceed

Prompt caching can lower costs when substantial portions of your input repeat across requests. OpenAI also offers other processing options and newer GPT models with different prices, so gpt-5 should not be treated as representing the price of every model carrying the GPT-5 family name.

For production applications, estimate costs using your actual average input and output sizes rather than only requests per month.

Practical checks before you proceed

Last reviewed: August 2026. Always verify OpenAI's official pricing before making financial decisions, because API models and prices can change.