by Corepass

Use Remy

Model selection and automatic routing

Choose a model, use Dynamic, and adjust reasoning settings.

Scope: The steps below use the desktop app.

Choose a model for your task, or select Dynamic to let Remy choose and move to a stronger model when the task gets stuck. Both options use your Remy account. You do not need a separate model-provider account or API key.

1. Choose how to work

What you need What to choose
You have a particular model in mind Select that model
You want Remy to choose a model for the task Select Dynamic
You want to adjust a model's reasoning settings Select a model that offers the setting, then adjust Effort

Check the model list in your account for available models and reasoning options. Dynamic does not display the specific model it selects.

2. Select a model yourself

  1. Open a chat in the project you want to work on and check the task you plan to send.
  2. Click the current model name beside the message box to open the model list.
  3. Select the model you want to use.
  4. Confirm that the selected model appears beside the message box, then send your task.

If you are unsure which model to use, you can start with Dynamic. To compare models, run separate tasks with the same source material and acceptance criteria. For tasks that change code, also start from the same project state.

3. Use Dynamic

  1. Open the model list before sending your task.
  2. Select Dynamic and check that it appears beside the message box.
  3. Describe the goal, relevant material, and completion criteria, then send the task.
  4. Review the output and add instructions if the result needs work.

Dynamic chooses a model for the task and can move to a stronger model if the task gets stuck. That change does not mean the problem has been solved. Review the actual result and any tests that ran.

You cannot currently switch models manually while a task is running. This is separate from Dynamic moving to a stronger model during execution. To select another model yourself, wait for the current execution to finish, then choose it before sending the next task.

4. Adjust reasoning settings

Some models offer an Effort setting.

  1. Open the model list and hover over the model to view its details.
  2. If Effort appears, open the options and select a level.
  3. Check the model and setting before sending your task.

Available levels can differ between models. If Effort is not shown, use the model's default behavior. When trying a model for the first time, you can keep the current setting and decide whether to adjust it after reviewing the result.

5. Common questions

Question What to do
I cannot find a particular model Check the active account and its model list. A screenshot from another account does not establish what your account can use
Dynamic does not show the model it selected This is the current display behavior. Judge completion from the task result and actual checks
I want to switch models during a task Wait for the current execution to finish, then choose the model for the next task
A stronger model still gives the wrong result Explain what happened, what you expected, and how to reproduce the issue, then ask for a correction. Model choice does not replace clear requirements and result checking

6. Keep exploring