Yes and No.
The Pentium, like the 486, came out first, followed by the Celeron and the 486-S, respectively. Both were intended for the lower performance requirements, that were driven by cost sensitivity. The cost and reduced performance were achieved in one fell swoop by multiplexing the data and address busses so that external accesses took twice as long, but in only half the hardware. Likewise the reduction in L2 cache to reduce die size reduces the manufacturing cost.
BUT, the instruction set is the same, otherwise, Mathcad and other high performance applications would crash and burn on a Celeron, but they don't, at least, not for an instruction set difference.
Changing the instruction set actually increases manufacturing AND support costs, since you now need different test programs for each chip, different compilers, etc. Additionally, the microcode is actually one of the smallest areas of the chip and reducing the instruction set does not really save that much die area.
Even TTFN