Use 60% Less Fable Tokens With This Orchestrator Setup

In this guide, you will learn two ways to stop burning Fable tokens on grunt work in Claude. Set Fable up as the planner and reviewer, then send browsing, coding, and research to Codex or a lower-cost Claude model so premium tokens are saved for the calls that need stronger judgment.

• Developers and builders who want Fable's judgment on a project without spending premium tokens on every file read, retry, and terminal task.
• Operators and researchers who use Claude for long, tool-heavy work and need a cleaner way to separate planning from execution.
• Anyone with a Claude paid plan who wants a worker to keep moving on routine work and call Fable only when the decision is genuinely hard.

You will build a simple model-routing setup: Fable owns the plan, risky calls, and final review, while a cheaper worker handles the token-heavy loop. The result is a self-checking AI worker that knows when to escalate instead of pulling Fable into every step.

• A Claude paid plan and the Claude desktop app
• Claude Code, updated to the current version
• A ChatGPT plan for the Codex-plugin route, or a lower-cost Claude model for the /advisor route

• OpenAI Developers plugin for Claude Code (https://developers.openai.com/learn/developers-codex-plugin)
• CJ Zafir on using Fable as the orchestrator and GPT/Codex as the executor (https://x.com/cjzafir/status/2074875092090470469)
• Peter Steinberger on using Fable as the advisor and executor models as workhorses (https://x.com/steipete/status/2074638582418231495)
• Oikon on Claude Code's /advisor command (https://x.com/oikon48/status/2074770258804769187)
• Anthropic's advisor tool documentation (https://platform.claude.com/docs/en/agents-and-tools/tool-use/advisor-tool)
• Anthropic pricing (https://platform.claude.com/docs/en/about-claude/pricing)
• Anthropic's effort documentation (https://platform.claude.com/docs/en/build-with-claude/effort)
• The guide brief (https://app.notion.com/p/395b5196e45b8170bcdbf9612821da77)

Save the delegation instruction as a Claude Code project instruction or reusable snippet. Then make a few worker templates for the work you repeat: a repo change, a research brief, a QA pass, or a content draft. Keep the same routing rule in each one: Fable plans, judges risk, and reviews the handoff; the worker does the bulk execution.

For a quick, low-risk task, skip the orchestrator setup and use the cheaper model directly. Bring Fable in when the work needs stronger judgment, not simply because it needs more tokens.