I made one call to chatgpt api with model gpt-3.5-turbo
in response it gave me usage information:
“usage”: {
“prompt_tokens”: 22,
“completion_tokens”: 35,
“total_tokens”: 57
},
so I can count usage in USD - it’s something like 0.00009200000204145908 if I follow price for “model”: “gpt-3.5-turbo-0613” in response
But via https://platform.openai.com/usage I can export usages with procing
and it gave me pretty different value:
1703808000.0,usd,0.0,0.0,0.0103,0.0,0.0,0.0,0.0,0.0,0.0
0.0103 seems too much for single request with “total_tokens”: 57
How come?
Can someone help me to figure out? Are there extra charges applied or just incorrect output for export at platform page?
I have an AI handy to ask…
“prompt_tokens”: 22,
“completion_tokens”: 35.prompt: $0.0015 / 1K tokens, completion: $0.0020 / 1K tokens
Please use code tools to calculate the exact price of this API call, and display the very small number in dollars without exponent.
=
The exact price of the API call, displayed in dollars without exponent, is $0.0001030000
The values that draw the graph are in pennies?
haha, exponent… that means export of data at openai platform page works incorrectly - there is no exponent in the csv file
Mentioning exponent is important knowing how code interpreter works. Small unformatted numbers by Python will quickly be exponentiated, and 2.6e-05 dollars isn’t so useful a unit visually.