Codex can manually enable a million contexts for GPT-5.6 Sol, making extremely long inputs more costly.
Codex officials have released a manual configuration method for the GPT-5.6 Sol million-token context, but this is not the default setting; inputs outside the default window will result in higher limits or API costs.
Thibault Sottiaux, head of OpenAI Codex, has announced a method for manually enabling a million-token context for GPT-5.6 Sol. The official OpenAI model page shows that the model's context window is 1,050,000 tokens, with a maximum output of 128,000 tokens.
This capability requires users to manually modify their Codex configuration and does not mean that all Codex sessions have been upgraded by default. OpenAI's documentation also reminds users that the default context has been tuned for common tasks; tokens exceeding the default window will consume usage quota more quickly. The API documentation also indicates that when more than 272K tokens are entered, the entire request will incur a higher fee.
Information Boundaries:A more accurate statement is "Codex can manually enable one million contexts for compatible models", rather than "All Codex models support one million contexts by default".
refer to:Codex Manager Configuration Guide、OpenAI Model Documentation、Instructions for using credit limit。