What is Mixture of Experts? Plain-English Definition

In more detail

A mixture-of-experts model is built like a clinic full of specialists rather than one overworked generalist. Inside, the model contains many smaller expert networks, and for each piece of your question a quick internal router wakes up only the few experts that are actually relevant.

The payoff is speed and cost: the model can be enormous in total knowledge while using only a fraction of itself for any single answer. It’s a big part of how today’s frontier models manage to be both very smart and fast enough to hold a conversation.

📌 See it in action

You ask one question about French cooking and another about computer code. Inside a mixture-of-experts model, different specialists switch on for each — while the rest stay dormant, like on-call doctors who only come into the hospital when their specialty is needed.

Goes with

Parameters →Frontier Model →Inference →

Now put the vocabulary to work.

Meet the models →Which AI is mine?