In additively manufactured (AM) Ti-6Al-4V, the role of martensitic α′ in governing brittleness versus toughness remains debated, largely because complex thermal histories and other intertwined physical factors complicate interpretation. To isolate and clarify the intrinsic effect of cooling rate, we employed a Gleeble thermal simulator, which enables precisely controllable cooling rates while simultaneously achieving ultra-fast quenching comparable to AM (up to ~7000 °C/s). By varying the cooling rate only, three distinct microstructures were obtained: α/β, α_m/α′, and fully α′. Compression tests revealed that the ultra-fast-cooled fully martensitic Ti-6Al-4V attained both higher strength and larger fracture strain, with densely distributed elongated dimples indicative of ductile failure. Three-dimensional microstructures reconstructed from microscopy, analyzed using an EVP-FFT crystal plasticity model, demonstrated that refined α′ laths and abundant high-angle boundaries promote more homogeneous strain partitioning and reduce stress triaxiality, thereby delaying fracture. These results provide potential evidence that extreme-rate martensitic transformation can overcome the conventional strength–ductility trade-off in Ti-6Al-4V, offering a new paradigm for processing titanium alloys and AM components with superior performance.

The laser-directed energy deposition (L-DED) process involves complex thermal and material interactions, posing challenges for controlling microstructural evolution in metal additive manufacturing. To enable rapid, efficient process optimization, this study develops a physics-informed mechanism-data fusion framework that couples macroscopic finite element method (FEM) and microscopic phase-field method (PFM) with a dual-level long short-term memory (LSTM) neural network. A multiscale physical model is employed to investigate grain growth behavior and the influence of process parameters on microstructural morphology. To map this non-linear temporal evolution, a dual-level data-driven surrogate model is developed. Compared to a conventional multilayer perceptron (MLP), the LSTM effectively captures time-dependent thermal accumulation. In recursive predictions, four initial input layers were identified as the optimal initialization length, enabling stable prediction and mitigating potential layer-wise error propagation, maintaining average R2 values of around 0.95. Using the average grain area as the primary microstructural control target, and incorporating the average grain width, a parameter range of 500–550 W and 13–14 mm/s was identified for achieving near-equiaxed β grains. The proposed framework is implemented as a modular computational pipeline that integrates finite element, phase-field, and surrogate models for reusable and extensible process-microstructure prediction in metal additive manufacturing.