SantageAI Glossary › Context Window
AI Glossary

What is Context Window?

A context window is the maximum amount of text an AI language model can process in a single interaction, measured in tokens.

What is the core idea behind context windows?

The context window is the model's working memory.

How do context windows differ from related concepts?

ConceptDifference
Context Window vs MemoryContext windows are per-session. Memory persists across sessions
Context Window vs KnowledgeKnowledge is from training. Context is from the current interaction
Context Window vs Token LimitToken limit includes both input and output within the context window

How do context windows work?

What are the limitations of context windows?

Why are context windows important?

Context window size determines what tasks an AI model can handle in a single interaction, from short questions to analysing entire documents or codebases.

How are context windows used in practice?

Context window sizes have grown dramatically. GPT-4 supports 128K tokens. Claude supports up to 200K tokens. Google Gemini supports up to 1 million tokens.

Frequently Asked Questions

Does a larger context window mean better performance?
Not always. Research has shown that models can struggle to use information in the middle of very long contexts effectively, a phenomenon known as 'lost in the middle.'
What happens when input exceeds the context window?
Content that exceeds the window is typically truncated, meaning the model cannot see or use that information in its response.