01 / The Model
Inference
Running the model to get an answer. Training builds it; inference uses it. Nothing about the model changes when you use it, which is why it cannot learn from your conversation.
How you’ll hear it
“Is it expensive to train our own?”
“We are not training anything. We pay for inference, each time we ask it something.”
Avoid
Confusing using a model with training one.
The longer version
Inference is the everyday act of sending a request and getting an answer, and it is what you pay for per use. Nothing about the model changes during inference, which is why it cannot learn from a conversation unless something outside it saves and reuses what happened.