Why do you use OPUS
I use it as a check occasionally. I do so much static that OPUS can't handle so it's not a very useful tool for me.
What are you using for a Base unit elevation to compare into an NGS benchmark to within "hundredths"? I've certainly never seen that using a "Here" base. Frankly I've never seen that using another NGS benchmark or an OPUS based elevation.
If we need a solid elevation reference we find a nearby NGS monument and typically our single 3 minute observation checks out within hundredths of the published elevation.
I would really, really love to see some of those checks, because I am highly skeptical of that claim.
EDIT: I should have said that it would be unusual, in my experience, to check in to 95% of NGS passive monuments within even a tenth or two, especially utilizing RTK methods. Even the BMs set during the NAVD88 levelling campaigns, and held fixed in the current hybrid geoid models, are rarely found to be within that range of published.
A super simple workflow for putting your typical project control on the current realization of the NAD83 datum is to set up your base to collect data for OPUS while you conduct your RTK ties. Presumably you will be set up for an hour of two. Send the base data into OPUS, get a solution, and shift your whole set of points onto that coordinate. Presto - you have the whole lot on a correct geodetic basis. That is at least 98% of what I use OPUS for.
As for just tying to a nearby NGS mon instead - apart from the time involved - there are exactly 4 such monuments within 10 miles of my city. One is completely unusable and two others are less than optimal for GPS due to the trees that have grown up around them. The last is a solid 10 miles away. Not a practical everyday solution.
For larger control projects I would make an effort to include those NGS monuments in my program but I would still weight OPUS more heavily in the adjustment. The CORS network OPUS is tied to is actively maintained and adjusted. Those passive marks were last tied 2 decades ago, and while they appear to be OK, it is hard to be certain that they are still in exactly the same position they purport to be.
After my recent experiences using OPUS, I find myself asking that question a lot