Astra and Fable are powerful models. Great! But can they do your work? And how well? Our solution is to create a personal benchmark: A series of tests that measure how well a model handles specific parts of your work. As part of our grand mission to free up our colleagues’ time, we’re building a benchmark for each one. @hammer_mt’s benchmark, for example, centers on creating slide decks. In @nityeshaga's testing, GPT-5.6 Luna outperformed Fable and GPT-5.6 Sol for Mike’s day-to-day…