Grok 4.5 Hits the Pareto Frontier on Coding — But Loses the Top Spot to a Startup Called Fable

xAI's latest model posts 51.2% on the APEX-SWE benchmark and a 30-point year-over-year jump, yet Musk himself acknowledged Grok placed second on real-world software engineering.

xAI's Grok 4.5 has closed most of the gap with the frontier — and, in an unusually candid admission, Elon Musk conceded it did not close all of it. "Grok places second after Fable on real-world software engineering," Musk wrote, naming a competitor rather than spinning the result. For a company whose founder rarely undersells, second place is a notable framing.

Unlock the full briefing

Get every story in today's briefing, the full archive, and the daily AI intelligence brief.

All stories today

Full archive

Daily brief

Cancel anytime. Payments powered by Stripe.