What amazes me is how difficult they do about what a product cost. With some trouble I found the 5720A, way above my head or what I want to spend on it. Didn't found 5725A. Did found that the 6.5 digit is no longer in catalog. They have now only a 5.5 digit Bench meter. Back to my question, if it is not possible to calibrate your own equipment, then what is the point in having, say, a 6.5 digit bench multi meter?. What I looking at?
Calibration is about having readings that are traceable to national standards. For many applications you need resolution (to determine how much a value changes with temperature, for example), but not necessarily accuracy. Also, good quality 6.5 digits like the classic HP/Agilent/Keysight 34401A are known to stay within their stated 1 year accuracy or most ranges for years, though if you want to certify this, you will need to have it calibrated every year. But for many hobbyists, knowing that a voltage is 5V +/- 1% but it's varying at most 0.5 mV over load or temperature is good enough.
If you need traceability, for example because your customers demand it, then you'd send it to a cal lab at the interval the manufacturer recommends (often 1 year, sometimes if you want higher accuracy). I haven't kept track of prices, but I'd think that calibration of a 6.5 digit meter would be in the ball park of $200.
The problem with calibrating it yourself is a) that you need another reference with traceable calibration with lower uncertainty than the desired uncertainty of the calibration and b) that you need stable sources for all signals that the meter needs, unless you want to do a limited calibration of only DCV and Ohms, let's say. For a full calibration might need DC voltages from 100 mV to 1000 V, resistance from 100 Ohm to 100 MOhm, DC currents from 10 mA to 2 A, AC voltages from 100 mV to 750 V from 10 Hz to 50 kHz and AC currents at 1 A and 2 A at 1 kHz (this is for the 34401A, check the service/calibration procedure for your DMM). Especially high frequency, high voltage AC is often difficult in your average lab. This all needs to have a stability much better than the desired uncertainty. So a basic 1/4W resistor will definitely not be stable enough for this comparison.
So either you would need a more accurate DMM that was calibrated before and stable sources for all these signals, or a calibrated multi-function calibrator that can source all these signs with a low enough uncertainty. This is how the cal labs do it, but they might be investing $100k in equipment and another $10k or so for yearly calibration. They make this back in the volume of equipment they calibrate.