Achromatopsia type 2 (ACHM2) is a severe, inherited eye disease caused by mutations in the CNGA3 gene encoding the α subunit of the cone photoreceptor cyclic nucleotide-gated (CNG) channel. Patients suffer from strongly impaired daylight vision, photophobia, nystagmus, and lack of color discrimination. We have previously shown in the Cnga3 knockout (KO) mouse model of ACHM2 that gene supplementation therapy is effective in rescuing cone function and morphology and delaying cone degeneration. In our preclinical approach, we use recombinant adeno-associated virus (AAV) vector-mediated gene transfer to express the murine Cnga3 gene under control of the mouse blue opsin promoter. Here, we provide novel data on the efficiency and permanence of such gene supplementation therapy in Cnga3 KO mice. Specifically, we compare the influence of two different AAV vector capsids, AAV2/5 (Y719F) and AAV2/8 (Y733F), on restoration of cone function, and assess the effect of age at time of treatment on the long-term outcome. The evaluation included in vivo analysis of retinal function using electroretinography (ERG) and immunohistochemical analysis of vector-driven Cnga3 transgene expression. We found that both vector capsid serotypes led to a comparable rescue of cone function over the observation period between 4 weeks and 3 months post treatment. In addition, a clear therapeutic effect was present in mice treated at 2 weeks of age as well as in mice treated at 3 months of age at the first assessment at 4 weeks after treatment. Importantly, the effect extended in both cases over the entire observation period of 12 months post treatment. However, the average ERG amplitude levels differed between the two groups, suggesting a role of the absolute age, or possibly, the associated state of the degeneration, on the achievable outcome. In summary, we found that the therapeutic time window of opportunity for AAV-mediated Cnga3 gene supplementation therapy in the Cnga3 KO mouse model extends at least to an age of 3 months, but is presumably limited by the condition, number and topographical distribution of remaining cones at the time of treatment. No impact of the choice of capsid on the therapeutic success was detected.