The transition to low-carbon energy systems results in strong interest for ammonia and hydrogen as alternative carbon-free fuels. Plasma-Assisted Combustion (PAC) offers a promising pathway to overcome their limitations in terms of combustion properties and pollutant emissions. However, incorporating non-equilibrium plasma effects into turbulent flame simulation remains computationally expensive: efficient and accurate modeling approaches are therefore essential. A Semi-Analytical (SA) model for PAC, named PACMIND-SA, enabling cost-effective prediction of discharge energy partitioning in ammonia–hydrogen–air mixtures, has been developed. Unlike some existing phenomenological approaches, PACMIND-SA does not require detailed time-resolved kinetic simulations for discharge energy branching evaluation, while detailed kinetics is only required once for the initial processes identification. Instead, it estimates the energy branching into chemical, thermal, and vibrational channels using electron-impact reaction data and a reaction rate-based analysis, informed by the electron energy distribution function. The model is validated against detailed plasma-kinetic simulations for various conditions, including single-pulse discharges in air, ignition of NH<inf>3</inf>-H<inf>2</inf> mixtures, NOx formation in various mixtures, MILD combustion, oxycombustion and plasma-assisted ammonia reforming. PACMIND-SA accurately reproduces key thermo-chemical trends such as ignition delays, radical formation, and pollutant emissions (NOx). In particular, the error on ignition delay time for a single pulse against the detailed mechanism is found between −9% and +6% over a wide range of E/N∈[100−500] Td and NH<inf>3</inf>-H<inf>2</inf> blending ratios (from pure ammonia to pure hydrogen). Tests performed on multiple pulses, 100 pulses at 20 kHz, showed a very good agreement with less than 1% error on the ignition delay time. Despite the assumption of constant thermochemical states during the pulse, the model shows excellent agreement across a wide range of discharge energies and fuel compositions. Additionally, the overall methodology can be applied to other type of fuel, and thus is not limited to the applied NH<inf>3</inf>-H<inf>2</inf> cases considered in this work, as shown in this work with CH<inf>4</inf>-air mixture. The low cost associated to the model parameter evaluation (O(ms)), three order of magnitudes lower than with the original PACMIND methodology (O(s)), could allow their on-the-fly evaluation in CFD solver, leveraging the need for lookup tables within a limited state space. Tools to generate the model coefficients are available at https://chemistry.cerfacs.fr/en/plasma/. Novelty and significance statement This work introduces a novel semi-analytic formulation for a phenomenological plasma-assisted combustion model that does not require detailed kinetic simulations any more once the main processes are identified. Instead, it incorporates the information from detailed chemical kinetic mechanisms directly, enabling efficient and cost-effective discharge energy branching evaluation. This approach significantly reduces computational overhead, facilitating integration into CFD solvers. Its generalizable framework allows for adaptation to different fuel/mixture compositions (ammonia, hydrogen and methane tested in this work), enhancing its practical use, provided that detailed kinetic mechanisms are available. This new model marks a step toward scalable and reliable modeling of plasma-assisted combustion.