For the OpenAI API , $100 buys different numbers of tokens depending on the model and whether you’re counting input or output tokens. Using the current published pricing, here’s the rough budget:

Model Input tokens for $100 Output tokens for $100
gpt-5 160,000,000 20,000,000
gpt-5-mini 800,000,000 100,000,000
gpt-5-nano 4,000,000,000 500,000,000
gpt-4.1 100,000,000 25,000,000
gpt-4.1-mini 500,000,000 125,000,000
These figures come from the current per-1M-token pricing: for example, gpt-5 costs $0.625 per 1M input tokens and $5.00 per 1M output tokens, so $100 covers about 160M input tokens or 20M output tokens.

Quick Scoop

A good rule of thumb is that 1 token is about 4 characters in English , or about ¾ of a word, so $100 can represent an enormous amount of text depending on the model.

Example

If you used gpt-5 and spent your whole $100 on input tokens, you could send about 160 million input tokens. If you spent it only on output tokens, you’d get about 20 million output tokens.

Important note

The exact count can vary by model, language, and whether tokens are input, cached input, or output, so the practical amount is always approximate.