This study presents a smart, low-cost, and portable gas sensing system based on a mid-infrared metasurface microspectrometer (MIMM) and machine learning (ML) algorithms. The MIMM, integrated with a commercial IR camera, is compact, lightweight, and consumable-free, measuring approximately 1 cm² and 1 g. The system can detect multiple gases, including carbon dioxide (CO₂), methane (CH₄), ammonia (NH₃), and methyl-ethyl-ketone (MEK), with high accuracy. It achieves 100% accuracy in detecting CO₂ and CH₄ at concentrations ranging from 10% to 100%, and 98.4% accuracy in detecting low concentrations of NH₃ and MEK. The system's performance is validated through experiments using a gas mixing setup and machine learning classifiers trained on collected data. The study demonstrates the feasibility of using IR spectroscopy combined with machine learning for smart, low-cost multigas sensing, making it suitable for applications such as air quality monitoring and hazardous gas detection.This study presents a smart, low-cost, and portable gas sensing system based on a mid-infrared metasurface microspectrometer (MIMM) and machine learning (ML) algorithms. The MIMM, integrated with a commercial IR camera, is compact, lightweight, and consumable-free, measuring approximately 1 cm² and 1 g. The system can detect multiple gases, including carbon dioxide (CO₂), methane (CH₄), ammonia (NH₃), and methyl-ethyl-ketone (MEK), with high accuracy. It achieves 100% accuracy in detecting CO₂ and CH₄ at concentrations ranging from 10% to 100%, and 98.4% accuracy in detecting low concentrations of NH₃ and MEK. The system's performance is validated through experiments using a gas mixing setup and machine learning classifiers trained on collected data. The study demonstrates the feasibility of using IR spectroscopy combined with machine learning for smart, low-cost multigas sensing, making it suitable for applications such as air quality monitoring and hazardous gas detection.