Skip to main content
All terms
Architectures

Long-Context Modeling

Techniques that let a model actually use and reason over very long inputs.

Definition

Long-context modeling covers the architecture changes, training methods, and serving tricks that help a model stay coherent and find the right details across very long inputs — big documents, whole codebases, or long chat histories. It goes beyond simply enlarging the context window to fixing the attention, memory, and retrieval problems that appear at that scale.