PSA: you don't need Fable for everything. Set it as your /advisor instead

Many users find Fable too costly or inefficient for all tasks. A common solution is to use Fable as an /advisor model, rather than the main model. This setup allows a primary model, such as Opus 5.5, to handle most work, while Fable is consulted only at critical junctures like before committing to an approach, when stuck, or before finalizing a task. This strategy leverages Fable's judgment without incurring its cost for every operation, optimizing both intelligence and expense.

This post offers a specific configuration for using Fable as an /advisor model, unlike general advice on model selection, to optimize cost and performance.