Benchmark wise 5% improvement over Sol in coding tasks and a 2-3% improvement over Fable 5.1 seems pretty disappointing, but maybe it is actually much better in

Benchmark wise 5% improvement over Sol in coding tasks and a 2-3% improvement over Fable 5.1 seems pretty disappointing, but maybe it is actually much better in real world usage. Let’s see