由于最近token吃的太猛了，就把不同角色都分配了合适的模型型号来缓解token压力，跟大家分享一下，欢迎大家指导我进步，其中说明了Codex的subagents多子代理是如何配置的，也有我目前使用的配置搭配： 

model = "gpt-5.4"
review_model = "gpt-5.4"
model_reasoning_effort = "xhigh"
plan_mode_reasoning_effort = "xhigh"
[agents].max_depth = 1
[agents].max_threads = 8
default = gpt-5.4-mini + high
worker = gpt-5.4-mini + high
explorer = gpt-5.4 + high 
codex-multi-agent-openai-config-guide.txt (11...