LLM cost per task, in price and in tokens
This free LLM cost per task calculator counts the tokens each task uses, then shows how much of your monthly API bill falls from a lower price per token and how much from fewer tokens per task.
Your task today
List price per million tokens: $2.00 input, $0.1000 cached input, $10.00 output. Read from the vendor pricing page on 2026-10-28.
System prompt, protocol and examples repeated on every request.
The part that changes with each request, such as a user record.
Hidden thinking tokens, billed at the output rate.
A model trained on your task
What remains of the repeated instructions once the rest is trained into the model. Capped at the instruction tokens in every prompt.
Changing the frontier model resets the two prices above unless you have edited them.
Your monthly cost
- OpenAI GPT-6.1 Sol
- $1,700.00
- $0.0170 per task
- Model trained on your task
- $500.00
- $0.005000 per task
Where the difference comes from, per month
- Lower price per token$0.00
- Fewer tokens per task$1,200.00
- Total difference$1,200.00 (70.6%)
Your costs come down on two fronts: a lower price per token than the large frontier models, and fewer tokens per task, since the instructions and context you repeat in every prompt get trained into the model itself. With your trained model at the frontier list rate and no prompt caching, the first line is zero and the saving comes from fewer tokens per task. A negative amount means that line costs more.
An estimate from list prices, as of October 2026.
Want the second line on your own traffic?
Ertas Endpoint serves a model trained on your work through one API. We measure the repeated share of your prompts from your own traffic before putting a number on it.
See how the endpoint works