TiO₂–carbon nanotube (TiO₂–CNT) hybrids are promising gas-sensor platforms because they combine the chemically active and photoresponsive surface of TiO₂ with the high conductivity and interfacial sensitivity of CNT networks. Yet cross-study comparison remains difficult because materials, device architectures, humidity, activation modes, and response definitions vary widely. This comprehensive review connects interfacial mechanisms and architectures with performance benchmarking and design rules. Direct gas-sensing evidence is distinguished from supporting interfacial studies and contextual analogues. We assess band alignment, oxygen vacancies, CNT functionalization, shell thickness, percolation, photoactivation, humidity, and analyte-specific behavior across NO_x, NH₃, H₂, H₂S, VOCs, LPG, O₂, and humidity sensing. The evidence shows that response is governed more strongly by interface continuity, spatial location, and the dominant current pathway than by nominal composition alone. Thin conformal coatings, discrete oxide decoration, and interpenetrating networks can each be advantageous; none is universally superior. Because response definitions and protocols remain inconsistent, we propose a benchmarking framework separating response magnitude from sensitivity, kinetics, selectivity, stability, humidity robustness, and activation requirements. These relationships yield practical design rules for controlling receptor chemistry, interface quality, percolation, activation energy, and validation conditions, providing a route from proof-of-concept demonstrations toward reproducible, low-power, application-relevant TiO₂–CNT gas sensors.