What Changed — Fable 5 Is Live in Claude Code
Claude Fable 5 returned to Claude Code on July 1, 2026, after a two-week export-control shutdown. It's now available in the model picker at exactly double Opus 4.8's price: $10 per million input tokens and $50 per million output tokens. On Every's Senior Engineer benchmark, it scored 91 against Opus 4.8's 63. But here's the catch—default to Fable 5 for everything and you'll drain your weekly limit before Thursday.
What It Means For You — The Routing Policy That Saves Money
The gap between Fable 5 and Opus 4.8 isn't uniform. It grows with task length and complexity. For a quick one-file edit or a bulk code reformat, Opus 4.8 or even Sonnet 5 ($2/$10) is more than adequate. But for long-horizon, multi-file, autonomous work—like refactoring a 50-million-line Ruby codebase (which Stripe did in a single day with Fable 5)—the premium pays off.
The key insight: Fable 5 draws down your weekly limit faster than Opus, and Opus already burns fast. So the real budgeting question isn't "$10 versus $5" per million tokens. It's "how many premium tasks fit in my week before I'm buying usage credits?"
Try It Now — Commands and Routing Strategy
How to switch to Fable 5
# Update first, or Fable 5 won't appear
claude update
# Switch mid-session (saved as your default afterward)
/model fable
# Or launch straight into it
claude --model claude-fable-5
My routing policy
Quick edit, single file Sonnet 5 ($2/$10) Cheap, fast, good enough Multi-file refactor, complex bug Opus 4.8 ($5/$25) Good balance for medium tasks Long autonomous run, migration, architecture Fable 5 ($10/$50) 91 vs 63 benchmark gap widens with task lengthThe /goal command trick
Fable 5's strength is judgment—it moves straight into implementation without asking permission. Combine it with Claude Code's /goal command to set a finish line the agent keeps working toward. This is where the model's lead really shows: it holds together over multi-hour runs with less steering.
What I Saw — Early Read
On one longer, multi-step task, Fable 5 held together better than Opus. It needed less steering to stay on track, and got closer to done without me babysitting each step. The way people are getting the most out of it is Claude Code's /goal command, which sets a finish line the agent keeps working toward instead of stopping short. That fits what I felt: give it your hardest, longest task, and the longer it runs, the more it pulls ahead.
Source: dev.to
