We present new methodological features and physical ingredients included in the 1D radiative transfer code HELIOS, improving the hemispheric two-stream formalism. We conduct a thorough intercomparison survey with several established forward models, including COOLTLUSTY, PHOENIX, and find satisfactory consistency with their results. Then, we explore the impact of (i) different groups of opacity sources, (ii) a stellar path length adjustment, and (iii) a scattering correction on self-consistently calculated atmospheric temperatures and planetary emission spectra. First, we observe that temperaturepressure (T-P) profiles are very sensitive to the opacities included, with metal oxides, hydrides, the alkali atoms (and ionized hydrogen) playing an important role for the absorption of shortwave radiation (in very hot surroundings). Moreover, if these species are sufficiently abundant, they are likely to induce non-monotonic T-P profiles. Second, without the stellar path length adjustment, the incoming stellar flux is significantly underestimated for zenith angles above 80 • , which somewhat affects the upper atmospheric temperatures and the planetary emission. Third, the scattering correction improves the accuracy of the computation of the reflected stellar light by ∼ 10%. We use HELIOS to calculate a grid of cloud-free atmospheres in radiative-convective equilibrium for self-luminous planets for a range of effective temperatures, surface gravities, metallicities, and C/O ratios, to be used by planetary evolution studies. Furthermore, we calculate dayside temperatures and secondary eclipse spectra for a sample of exoplanets for varying chemistry and heat redistribution. These results may be used to make predictions on the feasibility of atmospheric characterizations with future observations.