PURPOSE: This feasibility study examined a caregiver-implemented telehealth model of the Vocabulary Acquisition and Usage for Late Talkers (VAULT) protocol. We asked whether caregivers could reach fidelity on VAULT, if the protocol was socially and ecologically valid, and if late-talking toddlers could learn new words with this model. METHOD: Five late-talking monolingual and bilingual toddlers and four caregivers participated. The caregiver-related research questions involved measurements taken at multiple time points and replication across subjects but did not follow a specific research design. The toddler-related research questions included elements of a single-case design. Caregivers completed self-paced online training modules and then provided 8 weeks of VAULT to their children with remote coaching. Fidelity data were collected during coached sessions and through rating scales. Social and ecological validity data were collected via surveys and interviews. Children's word learning was measured before, during, and after treatment via production of targets and controls and via standardized vocabulary inventories. RESULTS: Caregivers demonstrated high fidelity to VAULT throughout treatment. They reported being comfortable with many aspects of VAULT. Feedback was mixed regarding the time required. Many reported their child was talking more as a result of the program. Visual analysis revealed that toddlers learned more target than control words, which was corroborated by Tau-U and d effect size analyses. CONCLUSION: A caregiver-implemented telehealth model of VAULT was feasible, was socially and ecologically valid, and benefited toddlers, making this a worthwhile model for future studies to examine. SUPPLEMENTAL MATERIAL: https://doi.org/10.23641/asha.21753872.