Read a mixture-of-experts model's active parameter count before you compare its cost

Is a model with more total parameters really the bigger, pricier one?

IntermediateRead 5 minDoing: about 45 minutes7 steps

▶ Start

What you'll have at the end

A corrected cost comparison of two real models, using active parameters instead of the total you started with

Before you start

You need

Two real mixture-of-experts models you are actually deciding between for a project, plus access to each one's own published model card and pricing page.

Not covered

Estimating cost for a dense model, where total and active parameters are the same number, or working out training cost rather than the per-token cost of running the model.

Leans on

7 steps

Where to go next

Your budget model keeps matching the flagship on real tasks

Checks whether the pricier model's extra capability is actually worth the corrected cost gap on a real task rather than on paper alone.

Checked 18 Aug 2026

Part of the Large Language Models (LLMs) cookbook