These days I decided to contact some calibration labs just because i wanted to have, at least for just one time, a professional evaluation of the status of my 34401A, the answer I got was basically "fuck you": none of the calibration labs near my home offer calibration for hobbysts...
I decided to do it on my own with my own procedure
I have one precision 0.03% decade resistor box/voltage divider (paid 60 euro on ebay) and one of those cheap 2.5, 5, 7.5, 10V chinese voltage references. My multimeter reads exactly the voltages reported on the reference box, and this gives me high confidence that the accuracy of at least the 10V range is spot on.
I have to start somewhere, and my assumption is that the 10V range, given the perfect readings with the voltage reference, is accurate.
I found out that many resistors on the decade are not as accurate as they should be, so i'm going to buy a precision 0.005% resistor for the most important ranges (100, 1k, 10k, 100k ohms), as well as a 4 wire short. With these i can calibrate the multimeter and repair and calibrate the decade box.
Ok, now i have a multimeter calibrated for voltage and resistance, and i also have a precise voltage source and a calibrated variable voltage divider. With these i can calibrate the current (at least in the mA range) and the lower voltage ranges. Be careful to account for the shunt resistance when doing the current measurement
Total cost: about 200 bucks, but you end up having extremely stable resistors you can always use as calibration standard, as well as a voltage reference to check if there has been some drift either in the multimeter or in the reference.
I don't care much for frequency and small AC voltages, so for that i have a 25MHz siglent function generator i won years ago just because i wrote a review about a cheap oscilloscope i bought online and Siglent decided to do a draft every 10 reviews to win the function generator (theyl little known and trying to advertise their products), basically the only thing i've ever won on my life. Soon i'll modify the function generator with a TCXO to get way better frequency stability.
Of course a metrologist would die at just the thought of following my procedure, but we have to work with very limited budget here, and this way, at lest for the most common ranges, you can be pretty sure your multimeter is well calibrated