In short
If you have a complex, multi-part task, it makes more sense to outsource the research and planning to a high-quality agency and the execution to a low-cost one. Everyone knows this.
How to Save Tokens in Claude Code 💡What not everyone knows is that Claude Code has a built-in feature for this: the /advisor command -> Run it once -> choose an advisor model (go with Fable 5 while it’s still available) -> Switch to Sonnet 5 right in the chat -> Profit How it works When a task is complex, Claude first calls the advisor, which provides a plan, and then the executor model follows that plan Anthropic: The Fable-advisor + Sonnet-executor combination delivers ~92% of Fable’s results at ~63% of the cost ✅When to use: — Long agent-based scenarios (coding, computer use, multi-step research) where most steps are mechanical but a good plan is crucial ❌When NOT to use: — simple tasks — Q&A — complex tasks where each step depends on the previous one (e.g., debugging)