Anti-PD1 (aPD1) immunotherapy transformed the clinical standard of care in melanoma and lung cancer, yet its efficacy is still variable among patients and tumor types. Deciphering the mechanisms underlying aPD1 responsiveness is limited when using static data, failing to capture the dynamics of migration, proliferation, and differentiation shaping tumor-immune interactions. To address this, we develop here a temporal single-cell and single tissue analysis of standardized treatment regimens and infer the dynamics of engineered tumor-specific CD8+ T cells. We discover that aPD1 treatment enhances activation of tumor-specific precursor cells to extend their cytotoxic window in the tumor microenvironment (TME), rather than re-activating mature, dysfunctional T cells. Elongation of the cytotoxic window by aPD1 is insufficient in patients or mouse models with limited influx of tumor specific early-effector T cells. Such influx is increased following treatment with an agonist for the co-stimulatory receptor 4-1BB, and combing this co-stimulation with aPD1 treatment yields strong synergistic efficacy in models unresponsive to aPD1 treatment. These results define that current checkpoint blockade strategies extend the cytotoxic window, but successful treatment may rely on co-stimulatory therapy for increasing the influx of tumor-specific precursor cells through this elongated window.