<p dir="ltr">Standard Model (SM) four-top quark production, in which two top-antitop pairs are produced in proton-proton collisions, is a rare process with strong potential to reveal new physics. This analysis targets the fully hadronic final state using proton collision data collected by the CMS detector during Run 2 and Run3 of the LHC, corresponding to 137 fb<sup>-1</sup> at √s = 13 TeV and 62 fb<sup>-1</sup> at √s = 13.6 TeV. To enhance signal sensitivity in the presence of large and complex backgrounds, a multi-step, data-driven approach is employed, incorporating boosted decision trees (BDTs) and deep neural networks (DNNs) for hadronic top tagging across a wide momentum range. Events are categorized based on the multiplicity of reconstructed top tags, and an event-level BDT discriminant is used to extract the signal. Backgrounds are estimated using control regions based on the ExtendedABCD method for the yield and Neural Autoregressive Flow-based ABCDnn method for the shape. The observed signal strength is measured to be μ = 2.7<sup>+1.1</sup><sub>-0.9</sub>, corresponding to an observed (expected) significance of 3.95 (1.65) σ and an observed (expected) limit on the cross section of 4.77 (0.98) σ<sub>t⊤⊤t</sub><sup>SM</sup>. A search for new physics in form of a top-philic resonance for a scalar color-singlet model yields an observed limit on the coupling strength of about 0.75 for resonance masses from about 600 GeV to 1600 GeV.</p>