Hello, I’m a novice in AI APIs. Does anybody know how many RAM is used in a PC when a Python program calls API with GPT-4-0125 or large-v2 whisper?
I want to know about the RAM usage, for use in automatization in some processes.
Hello, I’m a novice in AI APIs. Does anybody know how many RAM is used in a PC when a Python program calls API with GPT-4-0125 or large-v2 whisper?
I want to know about the RAM usage, for use in automatization in some processes.
Python memory usage is inconsequential. On a PC with windows 10, engaging with my command-line chatbot uses 39MB
. Less than the Windows terminal that runs it. 1/1000th of my system’s memory.
How much that grows will depend on how complex the code, and how many in-memory objects you have. Using Whisper, you will at least have the 25MB of the maximum audio file in memory, and if you do audio processing on it first, that can be much larger a memory footprint.
The OpenAI services all run on the server. You simply send a HTTPS request with the JSON text of the request (and binary audio file data), and receive back the text. Using urllib
network library calls instead of the OpenAI module is 11MB.