Background The efficacy on the Omicron variant of the approved early‐ coronavirus disease 2019 (COVID‐19) therapies, especially monoclonal antibodies, has been challenged by in vitro neutralization data, while data on in vivo antiviral activity are lacking. Materials and methods We assessed potential decrease from day1 to day7 viral load (VL) in nasopharyngeal swabs of outpatients receiving Sotrovimab, Molnupiravir, Remdesivir, or Nirmatrelvir/ritonavir for mild‐to‐moderate COVID‐19 due to sublineages BA.1 or BA.2, and average treatment effect (ATE) by weighted marginal linear regression models. Results A total of 521 patients [378 BA.1 (73%),143 (27%) BA.2] received treatments (Sotrovimab 202, Molnupiravir 117, Nirmatrelvir/ritonavir 84, and Remdesivir 118): median age 66 years, 90% vaccinated, median time from symptoms onset 3 days. Day1 mean viral load was 4.12 log2 (4.16 for BA.1 and 4.01 for BA.2). The adjusted analysis showed that Nirmatrelvir/ritonavir significantly reduced VL compared to all the other drugs, except vs. Molnupiravir in BA.2. Molnupiravir was superior to Remdesivir in both BA.1 and BA.2, and to Sotrovimab in BA.2. Sotrovimab had better activity than Remdesivir only against BA.1. Conclusions Nirmatrelvir/ritonavir showed the greatest antiviral activity against Omicron variant, comparable to Molnupiravir only in the BA.2 subgroup. VL decrease could be a valuable surrogate of drug activity in the context of the high prevalence of vaccinated people and low probability of hospital admission. This article is protected by copyright. All rights reserved.