No it's not: their model is only on GPT-4.5 level on a few, saturated/cherry-picked benchmarks.