Sixty X

Stripe reported that Claude Fable 5 compressed a codebase-wide migration from two months to one day. Fifty million lines of Ruby. A full-team effort reduced to a single engineer supervising an autonomous model. That is not a benchmark demo. That is production output on a live codebase at scale.

The model costs less than half what its predecessor did. Capability went up. Price came down. Both happened in the same release.

I keep watching how the companies already building with AI absorb these jumps. They don’t schedule evaluation committees. They slot new capability into existing workflows within days because they already have the operational muscle. Stripe didn’t get 60x because they have better engineers. They had a team and a process structured to put a new model to work the morning it shipped.

The organizations still debating adoption missed something worse than this model. They missed the reps that would let them use it on arrival. When the next release ships in three months, Stripe will absorb it again on day one. The companies still finishing their evaluation of this one will fall further behind — not by a percentage, but by a multiplier. That multiplier is the gap now.