You have two options, and I would avoid stack bond at all costs.

Option 1 - The CMU manufacturer should make a special corner block that incorporates an 8" wide end piece just for this purpose (at the corners).

Option 2 - You can substitute a standard 8" CMU at the corner located flush with the outside face of wall and then infill the gap with a solid 4" CMU.

For structural purposes and economy, avoid stack bond!! The only benefit of stack bond is the alignment of the cores. In some limited areas of the U.S. and the more advanced international masonry markets, core alignment may be automatically be obtained by using one core units with cross webs at 8" on center.

According to ACI 530, "stack bond" is referred to as "other than running bond". Running bond requires that the units lap by 1/4 of the units length.

For a 12" wall (assuming 16" long units), you can still avoid the techincal code provision imposed on stack bond by using all 12x8x16 units if you have alternating courses. The 4" difference between the width of a 12" block and the length of the 16" block below it provides the necessary lap. Some people do not appreaciate the appreance of a 4" "offset bond" even though it is not classified as a running bond.

As MarcbSE said, in most areas you can routinely get "L corners" in 6", 10" and 12" thickness that provide  a running bond appearance wising just one unit. This is the preferred detail.

If these units are not readily available, you can use an 8x8x16 unit at the corner and use 2 concrete brick to fill in the inside corner. The appearance is still the same as got get with the L corners, but you have the minor inconvenience od not doing it with just one unit. This requires no special units and be accomlished even in unsophisticated masonry markets.

For some architectural units and critical application, using the brick may be preferred or even specified.

concretemasonry,

There are only three things that I have found that make stack bond less desirable than running bond:

Can't use IBC empirical design
stack bond has a lower allowable shear strength when unreinforced
compressive area of concentrated loads get truncated at head joints, which means smaller areas for stack bond compared to running bond.

Are there any other reasons not to use stack bond?

twinnell -

The best explanation of one of the advantages is in the code commentary of ACI 530-02. (2.1.9  and the accompanying figure 2.1-14) since stress can only be transferred across the head joints laid in running bond. This has practical ramifications in arch action and crack control.

Probably the greatest advantage of running bond is economy and appearance. If appearance is a factor, stack bond is much more costly because it takes time and personal judgement to maintain the absolute uniformity and alignment of the units in a horizontal direction.

Both running bond and stack bond have vertical controls for the mason in terms of a level and a string line. For stack bond, each unit must be placed precisely (horizontally) to maintain the uniform width vertical head joint that must align exactly with the joint below. In running bond, a small variation in width is within the speifications and is not visably detected. Some architectural units can only be laid in stack bond.

It is very easy to maintain unit continuity vertical core alignment with running bond if the correct units are used.