Passive, heat actuated ejector pumps offer simple and energy-efficient options for a variety of end uses with no electrical input or moving parts. In an effort to obtain insights into ejector flow phenomena and to evaluate the effectiveness of commonly used computational and analytical tools in predicting these conditions, this study presents a set of shadowgraph images of flow inside a large-scale air ejector and compares them to both computational and first-principles-based analytical models of the same flow. The computational simulations used for comparison apply k-ε renormalization group (RNG) and k-ω shear stress transport (SST) turbulence models to two-dimensional (2D), locally refined rectangular meshes for ideal gas air flow. A complementary analytical model is constructed from first principles to approximate the ejector flow field. Results show that on-design ejector operation is predicted with reasonable accuracy, but accuracy with the same models is not adequate at off-design conditions. Exploration of local flow features shows that the k-ω SST model predicts the location of flow features, as well as global inlet mass flow rates, with greater accuracy. The first-principles model demonstrates a method for resolving the ejector flow field from relatively little visual data and shows the evolving importance of mixing, momentum, and heat exchange with the suction flow with distance from the motive nozzle exit. Such detailed global and local exploration of ejector flow helps guide the selection of appropriate turbulence models for future ejector design purposes, predicts locations of important flow phenomena, and allows for more efficient ejector design and operation.