Codex raises GPT-6.1 Sol's default context window to 500K tokens

OpenAI has raised the default context window for GPT-6.1 Sol in Codex and ChatGPT Work to 500K tokens, up from 272K, about 80% more. Tibo Sottiaux of OpenAI (@thsottiaux) announced it on X on Oct 11 as "Day 7" in a run of daily Codex updates.

WHAT CHANGED A Codex or Work session can now hold 500K tokens before it has to compact, which means summarising older context to free room. Sottiaux says the larger window cuts how often compaction happens and helps with tasks that need a lot of context kept in view. He says OpenAI will go higher in the future.

THE API WAS ALREADY BIGGER This is the default inside OpenAI's own apps. Through the API, GPT-6.1 Sol already lists a 1.05M-token context (922K input, 128K output) on models.dev and OpenRouter, at an unchanged $2 input and $10 output per 1M tokens.

WHAT WE DO NOT KNOW We could not yet read a matching entry on OpenAI's Codex changelog. The post does not say whether every plan gets the 500K default, and gives no date for the next increase.