Background and Context: This study looks at computational thinking (CT) assessment of programming artifacts within the context of CT integrated with science education through computational modeling. Objective: The goal is to explore methodologies for assessment of student-constructed computational models through two lenses: functionality and conceptual fluency. Method: This study uses data from research with EcoMOD, a 3rd grade ecosystem science curriculum in which student pairs program computational models of a beaver building a dam. Snapshots of programs for 47 student pairs collected over time are assessed. Findings: A functionality-based rubric provided assessment of student task progress, but was less successful at correctly identifying CT gains in programs that were partially correct. A rubric for conceptual fluency identified development of fluency in CT concepts of sequencing, loops, and conditionals. Implications: This study contributes to the literature by exploring affordances of different rubric-based CT assessments of student programs.