Abstract Present-day solar imaging spectrometers typically yield a few hundred million spectra in one hour of observing time. This number will increase by an order of magnitude for future instruments with larger 4k × 4k sensors, such as those planned to be used for the upcoming Daniel K. Inouye Solar Telescope. A fast quantitative analysis of such huge data volumes can be done by comparing the observations to an archive of pre-calculated synthetic spectra to infer the thermodynamic properties of the atmosphere. To analyze intensity spectra of the Ca ii IR line at 854 nm in the solar atmosphere, we generated an archive with 2,000,000 spectra under the assumption of non-local thermodynamic equilibrium (NLTE) with the NICOLE code. We tested its performance by inverting 60 spectral scans of Ca ii IR at 854 nm in the magnetically quiet Sun with 700,000 profiles each. Based on the inversion results obtained using the full archive, we constructed a smaller archive by keeping only the 70,000 archive profiles that were actually used. We can reproduce the observed intensity spectra to within a few percent using either the full or the small archive. For spectra with 30 wavelength points, this NLTE inversion approach takes 0.02 (0.35) s per profile to obtain a temperature stratification when using the small (full) archive, i.e., it can invert a single spectral scan in about 4 (68) hr. The code is able to simultaneously deal with an arbitrary number of spectral lines. This makes it a promising tool for deriving thermodynamic properties of the solar atmosphere from current or future solar high-resolution observations of photospheric and chromospheric lines.