Skip to main content

​
Short answer

review_model is the model Codex uses for code review workflows. model_reasoning_effort controls reasoning strength for models that support the Responses API. Both fields can affect quality, speed, and usage cost. Higher reasoning settings are better for complex tasks, but they can consume more.

​
Field relationships

FieldRole
modelDefault model for ordinary work
review_modelModel used for review workflows such as /review
model_reasoning_effortReasoning strength for the main model
/modelIn-session entry point for model or reasoning changes
/reviewEntry point for asking Codex to review current changes
model = "gpt-5.5"
review_model = "gpt-5.4"
model_reasoning_effort = "high"
For small quick edits, you can lower reasoning effort. For cross-file refactors, difficult debugging, or long-context tasks, higher reasoning effort is more appropriate.

​
Common mistakes

  • Thinking review_model replaces the main model for all tasks.
  • Treating reasoning effort as a temperature setting.
  • Using a model ID unsupported by the current provider.
  • Using the highest reasoning effort for every task.
  • Looking only at model names without checking actual usage and result quality.

​
About LLMEasy

Different models and reasoning settings affect usage. LLMEasy usage records help you compare input, output, and cost across real tasks. Copy model IDs from the GPT Key group so the current provider can recognize them.

​
References