arxivcs.AIcs.HC2026-07-06
CSTutorBench: Benchmarking Small Language Models as Tutors for Block-Based Programming
Large language models are increasingly explored as AI tutors, yet deploying them in K-12 settings raises concerns around privacy, cost, and reliance on proprietary models. Small language models (SLMs) offer a promising alternative, but selecting the right model for a specific edu…