Context. Among the blazar class, extreme blazars have exceptionally hard intrinsic X-ray/TeV spectra, and extreme peak energies in their spectral energy distribution (SED). Observational evidence suggests that the non-thermal emission from extreme blazars is typically non-variable. All these unique features present a challenging case for blazar emission models, especially regarding those sources with hard TeV spectra. Aims. We aim to explore the X-ray and GeV observational features of a variety of extreme blazars, including extreme-TeV, extreme-synchrotron (extreme-Syn), and regular high-frequency-peaked BL Lac objects (HBLs). Furthermore, we aim to test the applicability of various blazar emission models that could explain the very hard TeV spectra. Methods. We conducted a detailed spectral analysis of X-ray data collected with AstroSat and Swift-XRT, along with quasi-simultaneous γ-ray data from Fermi-LAT, for five sources: 1ES 0120+340, RGB J0710+591, 1ES 1101−232, 1ES 1741+196, and 1ES 2322−409. We took three approaches to modelling the SEDs: (1) a steady-state one-zone synchrotron-self-Compton (SSC) code, (2) another leptonic scenario of co-accelerated electrons and protons on multiple shocks applied to the extreme-TeV sources only (e–p co-acceleration scenario), and (3) a one-zone hadro-leptonic (ONEHALE) code. The latter code is used twice to explain the γ-ray emission process: proton synchrotron and synchrotron emission of secondary pairs. Results. Our X-ray analysis provides well-constrained estimates of the synchrotron peak energies for both 1ES0120+340 and 1ES1741+196. These findings categorise these latter objects as extreme-synchrotron sources, as they consistently exhibit peak energies above 1 keV in different flux states. The multi-epoch X-ray and GeV data reveal spectral and flux variabilities in RGB J0710+591 and 1ES 1741+196, even on timescales of days to weeks. As anticipated, the one-zone SSC model adequately reproduces the SEDs of regular HBLs but encounters difficulties in explaining the hardest TeV emission. Hadronic models offer a reasonable fit to the hard TeV spectrum, though with the trade-off of requiring extreme jet powers. On the other hand, the lepto-hadronic scenario faces additional challenges in fitting the GeV spectra of extreme-TeV sources. Finally, the e–p co-acceleration scenario naturally accounts for the observed hard electron distributions and effectively matches the hardest TeV spectrum of RGB J0710+591 and 1ES 1101−232.