I work with CD8 T cells and luminex assays, I usually work with atleast 100,000 cells per well in 200 uL final volume. Although luminex assays are very sensitive the kind of cytokine you want to detect is of prime importance. Are you stimulating the cells with CD3+28 ? or are they unstimulated? how long is stimulation?
Usually i can detect IL-2 as soon as 12 hr upon CD3+28 stimulation
If you are already doing the experiment and can only afford 2500 cells per well, you can give it a go. Just make sure that you use 96-well U-bottom plate (the cells like to be close to one another), and use negative and positive controls (add the same number of cells). Also use a smaller volume of media per well for example 100ul instead of 200ul. However, I think assay using 2500 cells is most likely not going to work.
I usually have 100,000 - 200,000 cells per 200ul (U-bottom).
Given the cost of luminex kits, I would first use an ELISA of a given cytokine that is significant to your experiment to assay the concentration obtained using different quantities of cells in a well. That would give you a concentration of cytokine for each cell quantity (and volume, conditions) that you could then relate to the standard curve of the Luminex kit. You could then choose the minimal quantity of cells (or an optimal volume or condition) that can still provide you with enough cytokine for the Luminex assay to be significant.