As tools become more sophisticated, computational modeling of biological systems is increasingly useful. Recent advances have allowed for the creation of executable models, ones that are expressed operationally using programs rather than denotationally as systems of differential equations. Expressing these models at programs has the advantage not only of being more compact, but also of being amenable to analyses developed for understanding programs. Our goal is to develop tools and techniques for executable knowledge, modeling biological systems by modeling mechanisms of cellular interaction. The vision of executable knowledge promises to transform areas from the design of biological experiments to the prescription of personalized drugs. Currently, a major obstacle is the difficulty of creating these models manually. There are two paths to generating models at scale: we can mine them automatically from the scientific literature, or we can generate them from higher-level specifications. For both paths there remains the question of how to navigate a space of possible models. As a solution I present ongoing work on Syndra, a deduction engine for reasoning across a space of possible biological models. Syndra integrates semantic, ontological reasoning with reasoning about the underlying biochemistry. I present Syndra's current support for reasoning in first-order logic, as well as its implementation using the Z3 SMT solver. I also discuss open questions with respect to extending Syndra for other kinds of reasoning.