Feature request for product/service
Chat
Describe the request
I currently do most of my work through an Agent that distributes tasks to Subagents. If the task list is too large or complex, the regular context window is not sufficient.
Expanding the GPT-5.4 context window to 1M completely solves this issue, but it creates a new problem: if GPT-5.4 launches a Subagent whose cost increases when using MAX mode (GPT-5.4, Sonnet 4.6), it may not be necessary in most cases, as the Subagents can handle the task in regular mode.
It would be great if it were possible to separately enable MAX for an Agent, but not enable it for Subagents.