Exploiting high-performance gas sensors is desirable for the on-site and accurate detection of drug precursor chemical gases. Here, the electron-proton conductivity metal-organic frameworks M3(HIB)2 were designed to discriminate typical drug precursor chemical gases. The strong d-π conjugation and substantial H2O ligands in M3(HIB)2 generate conducting pathways for electrons and protons, which contribute to novel gas-sensing properties. Remarkably, Fe3(HIB)2 demonstrates an ultrahigh response of over 379 toward 60 ppm of toluene at room temperature (RT). Furthermore, the adsorption/desorption behaviors of M3(HIB)2 can be tuned by systematically varying the metal center, causing distinctive gas sensing features for pattern recognition of drug precursor chemical gases. The recognition model was constructed using a convolutional neural networks-gated recurrent unit (CNN-GRU) algorithm, exhibiting a high recognition accuracy. The sensing mechanism is revealed by the Lewis and Brønsted acid site adsorption, due to competitive adsorption between H2O and analyte gases. This work paves the way for the development of proton-electron dual-conducting MOFs for high-performance gas sensors.