OpenAI Codex Context Window Reduced to 272k Tokens
July 19, 2026
OpenAI has decreased the Codex model context window from 372k to 272k tokens. This change affects the maximum amount of code and documentation that can be processed in a single inference pass.
HOW THIS AFFECTS YOU
●
builderYou must adjust your prompt engineering and RAG strategies to accommodate the smaller context window.
●
founderThis may impact the feasibility of processing large-scale codebases with your existing product architecture.