I'm playing with a TI op-amp, the THS4304. It's marked as a 3GHz op-amp on the datasheet's title page, but it has a 935 MHz GBW at G=5. Whatever, it is one of the faster VFB opamps you can get.
Its evaluation board comes loaded with parts for a non-inverting G=2 amplifier (two 249-ohm resistors). Both my scope and my pulse generator advertise a 0.8ns risetime. Setting up a short 250mV pulse, and looking separately at the pulse generator and amplifier outputs, I see a roughly 0.8ns risetime from either pathway. ;-) Here's a link to the scope screen shot.
The amplifier output appears largely identical to the input, but it's delayed about 0.7ns. That puzzles me, how does that delay come about? The time mismatch implies a fairly-large summing-junction error. Hmm, is that necessary to drive the op-amp's output slewing? 250mV in 0.8ns is about 310 V/us, and the op-amp is spec'd at 830 V/us (with a 500mV input error).
On a related topic, frequency-response poles carry with them signal time delays. But we know that Jim T often claims that a proper high-frequency op-amp model requires a delay element as well a poles and zeros. BTW, TI's SPICE model for the THS4304 does not have a delay element.