The Model's Rent on 1,500 Words Is About a Cent. Measure It Once, Then Quote by the Word

Why this pays

A writing or editing service that runs on a model pays rent by the token, and the rent is so small that the only mistake is not knowing the number. Both providers print it in the response to every call, so the way to price the service is not to read a rate card but to send one real job and read what comes back.

The per-word price you quote has the rent as one line and your time as the rest, and the rent will never move the total. What the measurement gives you is a floor you can prove, from a usage line you can show.

The play, in three lines

1. Send one job you actually sell through the API of Anthropic or OpenAI and write down the two numbers in the usage block, input_tokens and output_tokens.
2. Multiply by the published rates, divide by the words delivered, and write the per-word price list in the client's unit with that rent as one line inside it.
3. Save the job's input and output in your own folder, put a second provider's key beside them, and paste the ownership clause into your own client terms.

The path above walks every step. What follows is the reasoning behind each and the numbers you quote from.

Where the money comes from

• The client, in their unit. A word count, a page, a piece, an hour. Nobody buys tokens.
• Your time inside that price. Reading the result, checking the facts, making it sound like the client. Price it the way an editor prices it.
• Not the provider. Anthropic sells prepaid credits with no published minimum; OpenAI bills per token from the first call with no plan fee. Neither charges anything to keep an account open.

The pain you are selling the cure for

The client wants the words back tomorrow, in their voice, checked, and does not want to learn a tool to get them. The service sells the tomorrow and the checking. On your side the pain is the quote: a per-word price guessed from a rate card is either too high to win or too low to survive, and the fix is one measured job.

Step one: one real job, and the usage line

• Anthropic's first call: a POST to /v1/messages with a model, max_tokens and a messages array. Near the bottom of the response is a usage block with input_tokens and output_tokens.
• OpenAI's Responses API takes the same job as instructions and input and bills per token from the first call; pin a dated model snapshot for consistent behaviour.
• Send the job you actually sell: an edit of a 1,500-word draft, twenty product descriptions, a transcript summary. Three real inputs, three usage lines, and the numbers are close enough to price on.

Step two: tokens to cents to words

• Haiku 4.5: $1 per million input tokens and $5 per million output. Sonnet 5: $2 and $10, after the scheduled rise to $3 and $15 was cancelled. gpt-4o-mini: $0.15 and $0.60.
• Input tokens times the input price, output tokens times the output price, both per million, divided by the words delivered.
• The article-cost page runs the same arithmetic through seven passes and next to a human rate; the AI job cost calculator does it for any length.

Step three: the price, in the client's unit

• The price list is in words, pages, hours or pieces, and the model's rent is one line inside each that will not move the total.
• What moves the total is your time. The drafting guide shows how to write the brief once so the reading time falls with every job.

Step four: the folder

• Anthropic's privacy centre: API inputs and outputs are automatically deleted within 30 days. The only lasting copy is the one you store.
• OpenAI's services agreement: on termination OpenAI deletes all Customer Content within thirty days and hands nothing back.
• A service whose drafts live only in a console pays twice for the same words on the day it needs them, and loses the record of what the client approved.

Who owns what the model writes

• Anthropic: the customer owns the Outputs and keeps all rights to Inputs; Anthropic assigns its interest in Outputs and may not train models on Customer Content.
• OpenAI: the customer owns all Output; OpenAI assigns to the customer all of its right, title and interest in it; the customer keeps ownership of its Input.
• That clause is what lets you sell the words on. Put it in your own client terms in the same words.

What ends it

• Anthropic may suspend all access over a suspected Usage Policy breach and disclaims liability for any loss of data or profits that causes.
• OpenAI may limit or suspend access if required by law or if the customer breaches the agreement or OpenAI Policies, without prior notice where necessary.
• Neither account is sellable on its own: Anthropic's contract needs its prior written consent to assign, and OpenAI's is not assignable except to a successor of substantially all the business.
• The folder from step four plus a second provider's key means a suspension is an afternoon, not a business. The client-data page reads what each keeps of the conversations.

What to do today

1. Send one job you already sell through the API and read the usage block.
2. Multiply by the published rates and divide by the words.
3. Write the per-word price list with that rent as one line, save the input and output to a folder, and put the second provider's key beside them.