I Pushed Claude Fable 5 Hard for Ten Days

Anthropic's Claude Fable 5, the company's most powerful AI model, will lose free access on paid plans after July 12, shifting to API pricing at $10 per million input tokens and $50 per million output tokens. A user reports ten days of heavy coding use, finding the model's main limitation was not safety classifiers but other unstated barriers.

Member-only story I Pushed Claude Fable 5 Hard for Ten Days The wall I hit wasn’t the safety classifiers everyone worried about — and if you’re on a paid plan, you have about a day to learn from this for free. When Fable 5 came back on July 1 after its export-control suspension, I wrote an explainer on what it is and promised to report back once I’d actually used it in anger. This is that report — ten days of pushing Anthropic’s most powerful model on real, long, multi-step coding work, right through the window where it’s been included in paid subscriptions. The timing of this piece matters more than usual, so let me put it first: included Fable 5 access on Pro, Max, Team, and premium Enterprise plans ends July 12 at 11:59:59 PM PT — extended once already from the original July 7 cutoff. From July 13, using it means prepaid usage credits at full API rates: $10 per million input tokens, $50 per million output — exactly double Opus 4.8, and the most expensive pricing Anthropic has ever listed for a generally available model. So here’s what ten days of heavy use actually taught me — about the model, and about the two walls I did and didn’t hit.