Fable 5.2 reached the top of the GDPx benchmark with a 16.4 percent score, driven by software engineering and chip design gains. Exhaustive testing improves model accuracy, but it raises the average cost per task by 31 percent.
FireBrief
FireBrief