Model Serving & Economics
Model serving
Making a trained model available to applications so they can send inputs and receive outputs.
Example
A serving system keeps a trained model available and turns requests into predictions.
Why it matters
It lets a finished AI model support a working product that people can actually use.
You'll hear it
“The model works in testing; who's keeping it running for customers?”
Can you control it?
Set by the team
The people building or running the AI choose this setup. An everyday user generally needs their help to change how this part works.
Common questions
- Does a model need training again before each reply?
- No. It normally uses what it has already learned, though the service may add current information.
- Can two products use the same model?
- Yes. Their instructions, access to information and surrounding features can still produce different experiences.
- Why might a working model become unavailable?
- The computers, network or service around it can fail even when the model itself has not changed.