I'll call the first one "parallel setup" and the second "series setup", and compare them to a single CE BJT.
The parallel setup will dissipate half as much power per BJT (cooler), will have about the same voltage drop, will require the same amount of base current, and will have increased input capacitance. If we neglect the slight increase in input capacitance you can see that it has twice the pull-down power, thus is twice as 'fast' in dropping the voltage at the collector when turned on. In practice there will be no difference since the load capacitance is negligible. Turning them ON for an extended amount of time may lead to current hogging and thermal runaway. The two transistors, even if matched, will have slight differences in Beta-value and thermal characteristics, leading to differences in current. Current heats the device, which increases Beta, completing a positive current feedback loop called thermal runaway. With parallel devices it is also called current hogging because as one of the devices takes on slightly more current, the others take slightly less, and can lead to cascade of failures among parallel components -- once one branch fails, the rest must take on an increased load, further increasing the chance or rate of failure. To fix this, use current limiters in each parallel branch; in a CE BJT arrangement it is called emitter degeneration: a resistor at the emitter, often bypassed with a capacitor for AC design.