W4-11: A high-confidence benchmark dataset for computational thermochemistry derived from first-principles W4 data

► W4 theory re-validated against Active Thermochemical Tables data, RMSD = 0.08 kcal/mol, 3σ-confidence interval = 1 kJ/mol. ► Used to generate W4-11 benchmark dataset of 140 small molecules, evaluated plethora of ab initio and DFT methods against it. ► Best functionals: nonhybrid (M06), hybrid meta-GGA (PW6B95, M06-2X, BMK, and B1B95); if barriers important (M06-2X and BMK). ► DHDFT performance similar to composite methods, if barriers and resilience to nondynamical correlation important then DSD-BLYP is preferred. ► Among composite ab initio methods, G4(MP2)-6X offers best price/performance.