Builder Reports Stronger Results On Hard Tasks Versus Opus 4.5

@natolambert @FireworksAI_HQ i admit my memory of opus 4.5 is hazy at this point, but it is succeeding fairly well on some tasks I think would have been hard for opus 4.5, though it's hard to say for sure without testing, which I suppose I could do.