This is most likely the issue that I’ve mentioned before - hinge-like behavior of these kinds of connections. Run a frequency analysis and you should see it. The same model provides reasonable reaction forces in Abaqus where this issue is not present (due to proper implementation of rotational DOFs, MPCs and beam elements). As a workaround, you could try with 3 beams per bolt as shown here: Simplified bolt connection with beam element and rigid constraints - #100 by FEAnalyst
But, to be honest, if you check other related posts on this forum (especially the recent long discussions about bolted joints), you will see that beams are not really recommended for bolt modeling in CalculiX due to various significant limitations. In fact, it’s pretty much always easier and better to just use simple solids (cylinders) instead.