Inference
The part of a model's parameters that is actually used to process each token.
The part of a model's parameters that is actually used to process each token. In a Mixture of Experts model, a router sends each token to only a few experts, so compute per token follows the active count while memory still has to hold every parameter. Mixtral 8x7B, for example, has 47B total parameters but uses only 13B active parameters per token. Names like Qwen3.6-35B-A3B carry both numbers: 35B total and 3B active.
Primary source: arxiv.org
Example
Mixtral 8x7B, for example, has 47B total parameters but uses only 13B active parameters per token.
Hands-on guides, comparisons, and tutorials that cover Inference.
FAQ
The part of a model's parameters that is actually used to process each token.
Active Parameters sits in the Inference part of the AI stack. Understanding it helps you make better decisions when building, debugging, and shipping AI features.
Developers Digest publishes tutorials and videos that cover Inference topics including Active Parameters. Check the blog and YouTube channel for hands-on walkthroughs.
Related
A model architecture that routes each input to a small subset of specialized sub-networks ("experts") rather than activating the entire model.
The core technique inside transformers that lets a model weigh the relevance of every token relative to every other token in a sequence.
State space models are sequence models that carry a fixed-size hidden state forward through the input, updating it with learned linear equations, rather than comparing every token to every other token like attention.

New tutorials, open-source projects, and deep dives on coding agents - delivered weekly.