sonnet 5.5 just mogged astra tested both models with same prompt at high reasoning and man results are surprising i don't know what anthropic is feeding their models but this is just crazy how a sonnet model is able to give better output then astra? they better release model