Fable Cost-Saving Trick: Strong Models for Judgment, Cheap Models for Execution
vista8 · x · 2026-07-07
Summarizing Simon Willison's cost-saving strategies for developing with Fable (also applicable to other models): 1) Add a rule in the Claude system prompt to let the model decide and select an appropriate low-power model for coding tasks running in a sub-agent; 2) Use Opus or Fable in the main loop for complex tasks like architecture design; 3) Let the loop autonomously call sub-agents for mechanical tasks like writing and modifying code, primarily using Sonnet and Haiku. The core idea is to separate judgment from execution: use strong models for judgment, and cheap models for execution.
Related event: Fable Uses Model Routing to Slash Token Costs(3 posts)→
More from coding & agent
- shadcn's /improve skill paired with Grok on SuperGrok Heavy for plowing through issues — shadcn · 2026-10-27
- Paradigm Launches Envoys: Your Claude Code and Codex on Your Research Graph, Free — tensorqt · 2026-10-27
- LangChain to demo the internal design agent its team uses daily, this Wednesday — LangChain · 2026-10-27
- AXL launches remote MCP server exposing 1,284 sales and marketing APIs to Claude and Codex — EdTech_Padre · 2026-10-27
- GitHub's Thomas Dohmke ships Entire, a Git forge built for agent-human collaboration — carrycooldude · 2026-10-27
- OmarchyTV hits GitHub: Omarchy-style TV launcher with AI TUI and MCP — alexcovo_eth · 2026-10-27