Hi Dawson, 

I reckon that calibration must be done everytime you measure a new sample batch. 

"For each assay, you have the choice to run a new calibration or to use the values from the previous calibration. As you first use the instrument, you perform a new calibration each time. As you become familiar with the assays, the instrument, your pipetting accuracy, and significant temperature fluctuations within your laboratory, you can determine the level of comfort you have using the calibration data stored from the last time the instrument was calibrated." Cubit 4 user manual

Thanks for your answers. To clarify my own question, one needs to calibrate every batch of working solution in order to quantify the small changes effected by differences in the ratio of buffer:reagent. So a single lab could calibrate a large stock and use it for multiple batches, but probably this should done at a max of 1x per day given the light sensitivity of the reagent. With new Qubit kits that have premixed buffer and reagent, I suppose the purpose of calibration is to control pipetting error from user to user.