According to CoinMeta, on the 7th day of its 28-day product sprint, OpenAI announced that the default context size for GPT-6.1 and SOL during tasks in Codex and ChatGPT has been increased from 272,000 token to 500,000, representing an increase of about 80%. This larger context size can reduce the frequency of automatic context compression during long tasks, allowing the models to retain more details from previous conversations and tasks. Users can still adjust this value higher or lower at their discretion, with a maximum support of 1,000,000 token. Additionally, OpenAI will no longer charge an additional fee for using long contexts for GPT-6.1 and SOL that have subscribed to the ChatGPT quota.