OpenAI announced that it has expanded the default context window capacity for GPT-6.1 Sol, Codex, and ChatGPT Work from 272k to 500k, representing an increase of nearly 80%. When users run GPT-6.1 Sol within their subscription limits, no additional charges for long contexts will be applied; instead, only the normal subscription limits will be consumed. API calls will continue to be billed according to the pricing table. This expansion helps reduce the need for context compression operations, allowing the model to retain more details when processing long tasks. It effectively addresses the issues previously raised by developers regarding time-consuming compression and the tendency to overlook early requirements. Notably, OpenAI attempted to expand the context window in July but rolled back the change due to excessively rapid consumption of usage limits. The policy of waiving additional charges does not mean that task execution will consume fewer subscription limits; repeatedly reading cached content will still accumulate and consume subscription limits. Currently, the default context windows for Claude Code's Opus 5.5 and Sonnet 5.5 have reached 1M, twice that of OpenAI's new default, and they also do not charge additional fees for long contexts, featuring a mature context management mechanism. OpenAI stated that it will continue to increase the default context window capacity in the future.