The superior flight agility of hummingbirds is partly enabled by their ability to modulate wing motion via a highly evolved musculoskeletal system, consisting of a tiny, three-link, approximately 7-degee-of-freedom forelimb and attached feathers. However, due to the limitations of in-vivo measurement of muscle activity during free flight, it is challenging to gain insights into the physical activity of the musculoskeletal system, and to develop relevant functional model. In this study, we used a novel method to develop functional models of the hummingbird musculoskeletal system solely based on noninvasive intrinsic measurements. Specifically, we synthesized three sources of existing data, including 1) Computational fluid dynamics (CFD) simulation data for estimating forces applied to the wing, 2) Wing skeletal model from ?CT scan, and 3) Skeletal kinematic data from X-ray videos. The synthesis was conducted by using optimization methods to estimate the model parameters of a collection of hypothesized muscle functional models. Based on the identified models, we are able to gain comprehensive understanding of the functions of Pectoralis and Supracoracoideus, the two power muscles that contribute to all degrees of freedom of the lumped wing motion (i.e., stroke, deviation and pitching), along with two other muscle groups that contribute to the wing deviation and pitching. In particular, we gained insights into the physical behavior, e.g., stress & strain profiles, elasticity, and work loops of the two power muscles. In addition, these models also reveal the key design traits of hummingbird musculoskeletal system that should be translated to robotic flight for achieving hummingbird-level agility.