Understanding `gpt-4o-mini-tts` Pricing (Input Characters → Cost)

API
1

The image shows pricing details for text and audio tokens used in transcription and speech generation models, with specific costs per 1 million tokens for input, output, and estimated cost per minute. (Captioned by AI)
The image shows pricing details for text and audio tokens used in transcription and speech generation models, with specific costs per 1 million tokens for input, output, and estimated cost per minute. (Captioned by AI)1604×948 80.8 KB

:red_question_mark: How exactly is pricing calculated for gpt-4o-mini-tts?

I’m trying to estimate how much I’m going to pay given a specific number of input characters, but the pricing details seem a bit unclear.

From the docs:

  • Input: $0.60 / 1M characters
  • Output: $12.00 / 1M audio tokens
  • Estimated cost: $0.015 / minute (but this is clearly just an estimate)

My questions:

  1. What counts as input characters ? Is it the sum of instructions + input text ?
  2. If output pricing is based on audio tokens, what is the link between character count / audio tokens (given that 0.015$ is an estimate and not a real pricing)?

My goal is to be able to plug in a number of input characters and estimate the real cost before I scale up usage.

Any clarification would be super appreciated!