Is 200% a way to say *3 quicker ?
The little 10% reasoning performance decrease seems worth it.
Keep in mind, the speed improvement doesn’t come from the model running any faster (it’s the exact same architecture as R1, after all) but from using less output tokens while still achieving very good results.
We need about three orders of magnitude more tests to make these numbers meaningful.
Anecdotally, I can say that my personal experience with the model is in line with what the benchmarks claim: It’s a bit smarter than R1, a bit faster than R1, much faster than R1-0528, but not quite as smart. (Faster meaning less output tokens). For me, it’s at a sweet spot and I use it as daily driver.