Tokens and Billing / TOKEN WIKI
output token
In one sentence
output token is the result usage generated by the model. Different services may also list specific usage incurred during inference.
Why it matters
Output unit price and generation length can significantly affect task costs, and long answers do not always lead to higher business value.
Practical considerations
sets a reasonable upper limit on output length, records the success rate of complete tasks, and checks how the vendor measures inference usage.
FAQ
Can limiting the output length reduce costs?
can limit the generation usage limit, but it also needs to take into account answer completeness and retry cost.
