I see two core criteria for how low any new treaty can go: Targeting and survivability.
Regardless of what international law says about targeting, a sharp reduction towards minimum strategic deterrence will result in targeting population centres – everything else would be wasted. Russia and the U.S. each have around 30 with more than 500k inhabitants, China has over 200 of them; and one step lower the U.S. has 260, Russia 160 over 100k. That gives an idea of the number of warheads (and also yields) – survivability requirements, number of launch platforms, launch vehicles and the MIRV aspect follow from there.

De-MIRVing will get strong opposition from both sides. The U.S.’ best strategic deterrence instrument are the SSBNs, but non-MIRV’d SLBMs dramatically reduce the deterrence value of that leg – at least as long as the Ohios are the platform – and put too much weight on the vulnerable Minutemen (which will need a more survivable, possibly Midgetman stlye, single-warhead replacement in the future). The Ohio replacement will also have to be much smaller (Virginia class based?) and preferably with increased hull numbers to be more survivable, but is still 15 years away, and even then will for sure not be MIRV-less. All that is not do-able for the U.S. without spending serious money.
And the Russians will not de-MIRV, as they have to compensate for their technological inferiority, their lower platform and launcher numbers, the apparent vulnerability of their too few SSBNs with land-based MIRV’d mobililty. But the Russians are in a way better prepared already for minimum strategic deterrence, as their land-based component scales much better than the sub-heavy U.S. There is one angle though that might open at least land-based MIRV’d ICBMs for a horse trade (see below).

Couple of other aspects:

Agree that it would be a good idea to relieve the Air Force from the offensive strategic deterrence mission, in case of a minimum deterrence setup. The contribution of the big bombers to offensive strategic deterrence is minimal, AND such a move would free them for a 24/7/365 *defensive* strategic deterrence patrol within the framework of strategic missile defence, that could be as launch platform for airborne boost-phase interceptors (NCADE, PAC-3, THAAD versions) and kinetic and electromagnetic strikes against enemy launch complexes. All in all probably more useful than using them for the offensive side of things.

That brings up the question whether to include missile defense into a treaty. I’m sure the Russians want to. Especially the multi-KV technology has the potential to make strategic BMD a factor to reckon with. An interesting angle here could be that a Midgetman-style single-warhead ICBM and a multi-KV BMD missile have around the same weight. Now, here the horse ready for trading: There might be a way to meet the Russians half way and convince them to go for single-warhead ICBMs only, which would of course result in one warhead = one missile and thus would automatically limit the number of launch vehicles (which is what they want from the U.S.). It might be that a treaty could include a provision, that a certain percentage of these single-warhead ICBMs can be configured as multi-KV interceptors. The exact balance of offensive and defensive configured missile would be determined by what makes the Russians feel good and is acceptable for the U.S.

Then it might be a good idea to revise the test ban treaty to allow a minimum of tests. Would stabilize a minimum strategic deterrence posture by creating confidence in the arsenal.

And finally one should think about the model of the 1922 naval treaty for limiting the total overall yield of the force (strat and tac), as one aspect of reductions besides warhead and launch vehicle numbers. I would not go into the question of tactical vs strategic warheads, as even a tactical warhead on a high precision delivery platform can be used strategically. Such a treaty model could be expanded to also include the other strategic powers, China and the EU.