Abstract: Gas detection technologies are essential tools in maintaining safety and environmental standards across various applications. Through advanced sensors and analytical techniques, these systems aim to quickly detect and classify molecular content in an env ironment, providing valuable insights for early warning and effective response to incidents. In this work, we present the development of miniaturized, multiplexed, and connected electronic nose (e-nose) based on impedance spectroscopy technology. Our pla tform has been tested and optimized to process electrical responses of 15 conductimetric cells, each cell is tuned using a drop-casted conducting polymer poly(3-hexylthiophene) and 14 different triflate salts. The recognition of various solvents vapor (a cetone, methanol, isopropanol, water, ethanol and blends of the last two at various concentrations) relies on a Deep-Convolutional Neural Network based on a back propagation algorithm with two hidden layers of 64 and 32 neurons respectively. The achieved experimental results show an effective classification for the e-nose data to discriminate the alcoholic blends by type and composition, with high classification accuracy (~96%).