Abstract: This work aims to develop a high-temperature inelastic constitutive formulation for 316H stainless steel in sodium-cooled fast reactor that accurately describes stress-strain response under service temperatures above its significant creep threshold (425 ℃) and under thermomechanical loading. From an engineering-oriented perspective that favours simplicity and implementability, a separate-flow inelastic framework was adopted, in which creep and plasticity were described by distinct kinetic laws yet remain coupled through internal state variables. To compensate for the inherent limitation of separate-flow models in representing creep-plasticity interaction, a creep-plasticity interaction factor was introduced into the back stress evolution, so that creep deformation modified kinematic hardening. A strain memory term was further embedded in the isotropic hardening law to capture the amplitude-dependent peak-stress variation observed under cyclic loading, whereby plastic strain history influenced subsequent hardening capacity. For the creep law, only primary and secondary regimes were considered, consistent with design requirements for reactor-class high-temperature components that prohibited entry into tertiary creep during service life. A piecewise Norton-Bailey creep constitutive model was therefore employed. Detailed theoretical derivations were presented for the constitutive equations, together with a coupled iterative solution scheme for the creep and plastic consistency multipliers, which enhanced numerical efficiency and stability in the nonlinear regime. Model parameters were calibrated and validated against uniaxial tests at 550 ℃ . The nonlinearity solution algorithm demonstrates robust convergence and stable evolution of inelastic strains. Further validation against high-temperature mechanical tests on notched structural components at 550 ℃ shows that simulated displacements and surface strains agree with measurements within a relative error of 15%. The proposed framework retains a separable structure that is transparent to engineers, while delivering sufficient fidelity for creep-fatigue interaction assessment. With additional experimental verification, the model is suitable for engineering analysis of high-temperature 316H components.