Anthropic released Fable 5.1 while Opus 5 was still new, making the company look on fire. Its recent commercial opened with an actual house on fire, so the joke is on us for not taking it seriously. Once more, the company claims its new model is outrageously exceptional. Yet vendor-drawn benchmark charts cannot tell us the true performance out in the field. They merely support the existing adage in the community: let the frontier model design the work, then let cheaper models grind.

Anthropic’s chart contains spectacular numbers. Fable 5.1 scored 52.6 percent on Terminal-Bench-Science, against 29 percent for Opus 5, and nearly doubled Fable 5 on AutomationBench. Third-party results from Artificial Analysis are more boring, if not expected: 66 on its Intelligence Index against 63 for Opus 5, 62 for Fable 5, and 61 for GPT-5.6 Sol. On its professional-work benchmarks, Fable’s lead over Opus on GDPval-AA falls within the confidence interval, while the models are effectively tied on AA-Briefcase. First place, yes. But at what cost?

The cost figures bring the new guy down to Earth. At maximum effort, Fable 5.1 cost Artificial Analysis $3.76 per task, 20 percent more than Fable 5 and 1.6 times Opus 5. Drop Fable 5.1 to xhigh and it retains 65 of those 66 points for $2.72. That makes a frontier-only workflow unsustainable. Where one smarter orchestrator and many cheaper workers can do the job, that arrangement is cheaper by a large margin.

I have not tested Fable 5.1 yet. Anthropic calls Fable 5.1 “available” on Pro, but only through usage credits charged on top of the subscription. Included access begins with the $100 Max plan, where Fable is capped at half the weekly limit. I may fast-forward my planned upgrade maneuver from my annual Pro subscription, but GPT’s $100 plan is still my workhorse. Not to mention the rumors of an Astra model on the horizon. Fable 5.1 alone does not justify another $100.

This arrives the day after I wrote about the Max 20x debacle. Anthropic advertises four times Max 5x’s session capacity, while historical figures cited in a proposed class-action complaint put the weekly increase at about 1.7 times. Now “available on Pro” means available for purchase. You don’t even need a Pro subscription to use it on a pay-as-you-go basis. These statements survive fact-checking by choosing the friendliest possible denominator. That is not transparency. It is a dark pattern with unusually good lawyers — plus, for some reason, added footage of a burning house.

Fable 5.1 may prove exceptional. Until I can use it, the release proves something narrower: the smartest model does not need to play solo. Buy as little frontier judgment as the job requires, execute cheaply, and do not let Anthropic’s pricing pitch cloud your judgment.