{"data":{"slug":"context-window","etype":"concept","name":"Context Window","cat":"Core Technical","subtitle":"The model's working memory — from 2K tokens to millions in five years.","summary":"The context window is how much a model can consider at once. GPT-3 held 2K tokens; Gemini holds a million; long-context serving is why the KV cache (the stored state for every token in context) dominates inference memory and why chips like Rubin CPX exist just for the 'prefill' of huge contexts. Long context competes with retrieval (RAG) as the way to give models knowledge.","attrs":null,"kf":null,"aliases":["context length","long context","KV cache"],"ticker":null,"exchange":null,"hq":null,"founded":null,"website":null,"status":null,"region":null,"asof":"2026-01-15","sources":[{"url":"https://arxiv.org/abs/2403.05530","label":"Gemini 1.5 report"}]},"meta":{"found":true,"count":1,"served_from":"site cache (no live upstream calls)","as_of":"2026-10-14T04:29:31+00:00","disclaimer":"Research and information only — not investment advice. Structured records carry basis labels (vendor spec / reported / estimate); market quotes are delayed/indicative.","attribution":"Records compiled from vendor documentation, official filings and reputable public reporting; market quotes via financial data providers. See /data-sources."}}