While it's true that differentiating in the normal classroom "can be effective" it's also show in studies that it rarely comes close to approaching the effectiveness and growth rate that pull out classes do for GT students. This is only looking at academic outcome, it doesn't address the social / emotional aspects of GT education, which also are much easier to address in a classroom full of peers.
Knowing numerous GT educators in numerous schools, they often tell me that they discourage the idea of normal classroom differentiation as opposed to pull out programs, however, administration prefers normal classroom differentiation as it doesn't create scheduling problems and it puts an additional teacher into a normal classroom where the GT teacher can model differentiation to the normal classroom teacher, "Team Teaching" as the administration likes to label it. The problem with "Team Teaching" is that there is no study that shows it to be anymore effective.
I had a meeting with the local school district's curriculum coordinator and superintendent a couple of years back. The district claims in their literature and website to be "Data driven and committed to best practice" I asked what data and or studies they used to to determine that differentiation in the normal class room was best practice for GT students, the only answer I got back was, "Well.......lots of school districts do it that way." When I persisted about the data or studies they used to base their decision on, they changed the subject. No surprise there.
Schools will often spin the method they use in a very pretty picture, however, when it comes to backing that method with data and studies, they have trouble doing so.
