Fable 5.1 medium effort matches Fable 5 high, no longer breaks prompt cache
lydiahallie · x · 2026-09-05
Lydiahallie shares hands-on observations of Fable 5.1: at medium effort it performs roughly on par with Fable 5 at high, making it the first model where medium is her default in Claude Code. She also notes an engineering fix — switching /effort levels no longer breaks the prompt cache — and recommends builders try /effort medium this weekend.
More from coding & agent
- Roasting your agent's plan for hours is a core skill of the agentic era — burny_tech · 2026-09-05
- Anthropic engineer's agent stores a map of what it doesn't know and checks it before answering — anirbanbandyo · 2026-09-05
- 300 AI agents replace a research team — the trick is collapsing outputs into one context graph — anirbanbandyo · 2026-09-05
- Yohei Nakajima's AI now outputs a trustworthy prioritized daily task list — RachelVT42 · 2026-09-05
- doodlestein to Test Google's Astra on His Most Ambitious Project, FrankenSim — doodlestein · 2026-09-05
- Demo Combines GPT-6, CustomUse MCP and Three.js Procedural Rigging — majidmanzarpour · 2026-09-05