VISTA (V-domain Ig suppressor of T cell activation) is an emerging immunological checkpoint receptor that inhibits T cell activation, which is crucial for tumor immune evasion. Blocking the VISTA signaling pathway has emerged as a viable cancer immunotherapy method since it improves antitumor immunity across a wide range of tumor types. However, the number of small-molecule inhibitors targeting VISTA remains limited. In this study, we used an integrative in silico approach to uncover new VISTA-targeted immunotherapeutic candidate. A machine learning classification model trained on known VISTA inhibitors was created to screen three specific libraries: the Immuno-Oncology Library, MPD3, and NPACT. The Random Forest (RF) algorithm outperformed other models, achieving high accuracy (0.99) and was used for virtual screening. The Random Forest (RF) model screening identified 2903 potential active compounds from the three libraries. Subsequent filtering using Lipinski's Rule of Five criteria yielded 166 drug-like compounds for further analysis. These compounds were evaluated for their interaction profile and binding conformation through molecular docking. Five compounds PubChem 11,669,392 (- 8.5 kcal/mol), PubChem 442,827 (- 7.7 kcal/mol), PubChem 9,866,696 (- 8.0 kcal/mol), Z432360790 (- 8.1 kcal/mol), and Z229513896 (- 7.5 kcal/mol) were selected based on their strong binding affinities, stable interactions with key VISTA residues such as Arg54, Arg127, His66, and Cys51, and favorable ADMET profile. Subsequently molecular dynamics simulations confirmed the structural stability of the ligand-protein complexes. MM-GBSA free energy analysis revealed high binding affinity, with PubChem 11,669,392, PubChem 442,827, and Z432360790 exhibiting binding energies of - 15.49, - 13.46, and - 16.35 kcal/mol, respectively. This study identified small molecules as potential immunotherapeutic targeting VISTA, which could contribute to the development of effective cancer treatments.