Reaction of 2-hydroxy3-methoxybenzaldehyde (o-vanillin) with 1,1,1-tris(aminomethyl)ethane, Me-C-(CH2NH2)(3), or with N,N ',N"-trimethylphosphorothioic trihydrazide, P(S)[NMe-NH2](3), yields two tripodal LH3 and (LH3)-H-1 ligands which are able to give cationic heterotrinuclear [LCoGdCoL](+) or [(LCoGdCoL1)-Co-1](+) complexes. The Co-II ions are coordinated to these deprotonated ligands in the inner N3O3 site, while the Gd-III ion is linked to three deprotonated phenoxo oxygen atoms of two anionic [LCo](-) or [(LCo)-Co-I](-) units. Air oxidation of these trinuclear complexes does not yield complexes associating Co-III and Gd-III ions. With the first ligand, the structurally characterized resulting complex is the neutral mononuclear LCoIII compound, while in the second case, oxidation of the Co-II ions turned out to be impossible. The [L(1)CoLnCoL(1)](+) complexes behave as single-molecule magnets with effective energy barriers for the reversal of magnetization varying from U-eff = 51.3 K, tau(o) = 2 X 10(-6) s for the yttrium complex to U-eff = 29.5, 29.4, 27.4 K and tau(o) = 1.3 X 10(-7), 1.47 X 10(-7), 1.50 X 10(-7) s for the gadolinium ones, depending on the used counteranions. The energy decrease is compensated by the suppression of quantum tunneling of magnetization in absence of applied field, thanks to the introduction of a ferromagnetic Co-Gd interaction. Current work also shows that uncritical use of conventional spin Hamiltonians, based on quenched orbital momenta, can be misleading and that ab initio calculations are indispensable for establishing the picture of real magnetic interaction. Ab initio calculations show that the Co-II sites in the investigated compounds possess large unquenched orbital moments due to the first-order spin-orbit coupling resulting in strongly axial magnetic anisotropy. Although the Co-II ions are not axial enough for showing slow relaxation of magnetization by themselves, blocking barriers of exchange type are obtained thanks to the exchange interaction with Gd-III ions.