The tsetse fly ( Glossina genus) is the main vector of African trypanosomes, which are protozoan parasites that cause human and animal African trypanosomiases in Sub-Saharan Africa. In the frame of the IAEA/FAO program ‘ Enhancing Vector Refractoriness to Trypanosome Infection’, in addition to the tsetse, the cereal weevil Sitophilus has been introduced as a comparative system with regards to immune interactions with endosymbionts. The cereal weevil is an agricultural pest that destroys a significant proportion of cereal stocks worldwide. Tsetse flies are associated with three symbiotic bacteria, the multifunctional obligate Wigglesworthia glossinidia, the facultative commensal Sodalis glossinidius and the parasitic Wolbachia. Cereal weevils house an obligatory nutritional symbiosis with the bacterium Sodalis pierantonius, and occasionally Wolbachia. Studying insect host-symbiont interactions is highly relevant both for understanding the evolution of symbiosis and for envisioning novel pest control strategies. In both insects, the long co-evolution between host and endosymbiont has led to a stringent integration of the host-bacteria partnership. These associations were facilitated by the development of specialized host traits, including symbiont-housing cells called bacteriocytes and specific immune features that enable both tolerance and control of the bacteria. In this review, we compare the tsetse and weevil model systems and compile the latest research findings regarding their biological and ecological similarities, how the immune system controls endosymbiont load and location, and how host-symbiont interactions impact developmental features including cuticle synthesis and immune system maturation. We focus mainly on the interactions between the obligate symbionts and their host’s immune systems, a central theme in both model systems. Finally, we highlight how parallel studies on cereal weevils and tsetse flies led to mutual discoveries and stimulated research on each model, creating a pivotal example of scientific improvement through comparison between relatively distant models.