Context. This paper investigates the star formation histories (SFHs) of a sample of massive galaxies (M_⋆ ≥ 10¹⁰ M_⊙) in the redshift range 1 < z < 4.5.

Methods. We analyzed spectro-photometric data, combining broadband photometry from HST and JWST with low-resolution grism spectroscopy from JWST/NIRISS, obtained as part of the MIRI Deep Imaging Survey program. SFHs were derived through spectral energy distribution fitting using two independent codes, BAGPIPES and synthesizer, under various SFH assumptions. This approach enables a comprehensive assessment of the biases introduced by different modeling choices.

Results. The inclusion of NIRISS spectroscopy, even with its low resolution, significantly improves constraints on key physical parameters, such as the mass-weighted stellar age (t_M) and formation redshift (z_form), by narrowing their posterior distributions. The massive galaxies in our sample exhibit rapid stellar mass assembly, forming 50% of their mass between 3 ≤ z ≤ 9. The highest inferred formation redshifts are compatible with elevated star formation efficiencies (ϵ) at early epochs. Nonparametric SFHs generally imply an earlier and slower mass assembly compared to parametric forms, highlighting the sensitivity of inferred formation timescales to the chosen SFH model–particularly for galaxies at z < 2. We find that quiescent galaxies are, on average, older (t_M ∼ 1.1 Gyr) and assembled more rapidly at earlier times than their star-forming counterparts. These findings support the “downsizing” scenario, in which more massive and passive systems form earlier and more efficiently.