Model Serving & Economics
Mixture of experts
MoE
A model architecture that routes each input through only some specialized sub-networks rather than the entire model.
Example
Only a few specialist components activate for each request in a mixture-of-experts model.
Why it matters
It can provide a larger range of learned skills without using every part for each reply.
You'll hear it
“How much of that model actually runs for one request?”
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
- Are the experts separate people?
- No. They are parts of the AI model, not human specialists checking the answer.
- Can I pick which expert answers me?
- Usually the model's internal setup makes that choice, rather than offering a user selection.
- Does this guarantee expertise in my industry?
- No. The name describes the design; useful industry knowledge still needs to be tested.