Detection performance comparison between this work and the detection limits computed in Vigan et al. (2021). Left: comparison between the sensitivity achieved by Vigan et al. (2021) using classical post-processing algorithms (Langlois et al. 2021) on the F150 (dashed lines) compared to the sensitivity achieved using PACO on the same sample of stars (plain lines). The lower limit seen around 100 au at 0.5 M_Jup is due to the COND model, which does not provide inputs for lower masses. Right: probability increase (${\mathcal{P}}_{\text{PACO}}$(sma, M_Jup) – ${\mathcal{P}}_{\text{SPECAL}}$(sma, M_Jup)) between the PACO analysis and the one presented in Vigan et al. (2021).