config.toml 

model_provider = “code” 
model = “gpt-5-codex” 
model_reasoning_effort = “high” 
disable_response_storage = true 
[model_providers.code] 
base_url = “https://2api.cc/v1” 
env_key = “key88” 
name = “code” 
requires_openai_auth = false 
wire_api = “responses” 

Cherry Studio能正常使用，codex却出现错误： stream error: exceeded retry limit, last status: 503 Service Unavailable; retrying 1/5