The first million years of the Solar System is a key period when the first solids were formed from the nebula gas. The chronology of the different processes at the origin of these solids is still largely unknown and relies strongly on the assumption made of homogeneous distribution for short-lived radioactive nuclides such as 26Al. This assumption is questioned. In this study, in situ 26Al–26Mg isotope systematics was studied with high precision in 12 calcium, aluminum-rich inclusions (CAIs) (1 type A, 2 type B, 5 type C, and 4 fine grained spinel-rich), 2 amoeboid olivine aggregates (AOAs), and 2 Al-rich chondrules from Efremovka and Vigarano. The (26Al/27Al)i in these early Solar System solids (the subscript ‘i’ stands for the initial isotope ratio obtained from the mineral 26Al isochron) range from ~ 6.5 x 10-5 ro 0.2 x 10-5 with δ26Mg1* from -0.08 to +0.37‰. The (26Al/27Al)i and δ26Mg1* of CAIs and chondrules can be explained by formation of their precursors from a homogeneous reservoir (Solar System Initial, noted hereafter SSI) with initial magnesium isotopic composition of δ26MgSSI* = -0.52 ± 0.013‰ and initial (26Al/27Al)SSI abundance of (5.62 ± 0.42) x 10-5. The high precision magnesium isotope data obtained in the present study and literature data allows identifying a few epochs of formation/reprocessing of CAIs. The time periods of these epochs correspond well with the median life times of the pre-main sequence evolution of stars of Solar mass if we anchor the (26AI/27AI)SSI = (5.62 ± 0.42) x 10-5 to the beginning of class I phase. This provides a natural explanation to the range of (26Al/27Al)i – (∼6 to 0.02) × 10−5 seen in corundum grains, FUN (Fractionation and Unidentified Nuclear Effects) CAIs, ultrarefractory CAIs, normal CAIs, and chondrules, and suggests a possible relationship between the astrophysical conditions and the formation of these early solids. Corundum grains, FUN CAIs, ultrarefractory CAIs would have formed during the class 0 of the pre main-sequence, during the collapse of the parent molecular cloud, before homogenization was complete while normal CAIs and chondrules formed later during class I to the beginning of class III. Alternatively, if one wants to explain the present data without calling for an early homogenization of Al and Mg isotopes in the forming Solar System, at least four distinct isotopic reservoirs are required to produce by mixing the range observed in δ26Mg1* and (26Al/27Al)i. There is no obvious astrophysical explanation of the existence of such distinct reservoirs, which makes this hypothesis quite ad hoc and thus very unlikely.