Planting design is a complex task that requires designers to consider a multitude of factors, many of which are ephemeral or transitory over the seasons or as a plant matures. Generative artificial intelligence (AI) may streamline the planting design process by introducing increased efficiency during the selection phase. This study examines the ability of GPT 3.5-based pplications to identify suitable plants for several different planting schemas and evaluates the reliability of the prompts both within and between samples. Results suggest that both ChatGPT and direct use of GPT 3.5 via API can be a valuable planting design resource, but that there may be significant bias in the results, given the type of model selected. Understanding and mitigating this bias will be important for landscape architects who seek to use ChatGPT or other GPT models via API in planting design.