Virtual tables cannot be shared across modeling sessions or in batch dataflows.
Unlike datasets, virtual tables are local to the user.
You have two options to share that data with others:
1) export the table so that it can be re-imported as a dataset (and all users can share that dataset)
2) create a batch dataflow and make it available for others to execute; they can parameterize the output to export to resultant table to a Xcalar table in their session. This technique is not possible for Modeling licenses of Xcalar because batch dataflow execution is not supported in a Modeling licensed Xcalar cluster.