Immunotherapies have revolutionized the treatment of multiple myeloma in the last decade. Despite these recent innovations, a significant subset of patients fail to respond or relapse after treatment. In addition, many of these immunotherapies lack biomarkers to determine preemptively whether a patient would likely respond to treatment. Many of these immunotherapies are designed to enhance the endogenous T cells of the patient to drive tumor control and regression however what these T cells are targeting on the tumor is largely unknown and the current consensus is these immunotherapies are broadly-acting on the T cell compartment. Instead, we hypothesize that immunotherapies are selectively activating tumor-specific CD8 T cells that underlie the clinical response in refractory myeloma patients. We performed high-throughput antigen screening and multi-omic single-cell analysis of T cells derived from two clinical cohorts of refractory myeloma patients being treated with immunotherapies with the aim of isolating tumor-reactive CD8 T cells. We then validated their antigen specificity and cytotoxic potential and tracked their single-cell phenotype within the tumor and blood over time. We find that relapsed patients lack neoantigen-specific CD8 T cells and that broad-based immunotherapies do not cause large scale changes in phenotype in the blood. However, we do find tumor-specific CD8 T cells targeting tumor-associated antigens in a subset of patients including a novel tumor-associated antigen target. We find tumor-specific CD8 T cells exhibit diverse phenotypes, that they all significantly expand during treatment, and that their phenotype and clonality help drive and explain clinical response. Lastly, we find that the TCR repertoire within the blood did not predict response to treatment, but the TCR clonality of the T cells infiltrating the tumor significantly predict the clinical response to immunotherapy suggesting the preexisting TCR repertoire within the tumor determines whether a patient will respond. We conclude that broad-based immunotherapies are not truly broadly acting, but rely on the activation and functionality of endogenous tumor-specific T cells to drive clinical response.