Context window · Token limit
The context window is the maximum amount of text a model can process at once, measured in tokens. It limits how many sources an AI engine can cross-check in one answer: everything retrieved must fit in the window along with the question and the answer.
Standalone definition · 45 words
How does the context window work?
The prompt, the retrieved passages and the answer being generated share the same limit. A RAG system therefore selects a few passages, not whole pages. Larger windows allow more of them, at higher cost and latency.
Does a large window guarantee better use?
Not necessarily. The “Lost in the Middle” study (Liu et al., 2023) shows that models use information placed in the middle of a long context less well than information at the start or end. More context does not mean more accuracy.
Consequences for content
Short, dense, standalone passages fit the window better and are easier to compare. Diluted content, with a lot of filler, consumes token budget without adding a citable fact.
Frequently asked questions
How big is a context window?
It varies by model and changes quickly. Refer to the provider’s documentation.
Is a token a word?
No, it is a word fragment. A French word averages more than one token.
Does the engine read my whole page?
Not necessarily: it selects passages according to the query.
Related terms
Further reading
- Checklist GEO and LLM / AI 2026
- 2026 GEO Tools Comparison: The Decision Guide
- ← Back to the GEO glossary (114 terms)