The field of artificial intelligence is moving towards the development of more efficient, customizable, and accessible AI-powered educational tools. Recent studies have explored the use of small language models (SLMs) as a viable alternative to large language models (LLMs) for tasks such as curriculum-based guidance, due to their lower computational and energy requirements. Additionally, there is a growing interest in the application of reinforcement learning to improve the performance of LLMs in various tasks, including conditional semantic textual similarity and persuasive price negotiation. The development of novel frameworks and architectures, such as Point-to-List Reinforcement Learning and Reward-Enhanced Policy Optimization, has shown promising results in these areas. Noteworthy papers include the introduction of PoLi-RL, a Point-to-List Reinforcement Learning framework that achieves state-of-the-art results in conditional semantic textual similarity, and the proposal of Reward-Enhanced Policy Optimization, a reinforcement learning post-training framework that aligns an LLM with heterogeneous rewards. Furthermore, researchers are also exploring the use of LLMs in math education, with studies showing the potential of LLMs to generate standards-aligned educational math word problems and improve student learning outcomes. Overall, the field is moving towards the development of more advanced, efficient, and accessible AI-powered educational tools that can improve learning outcomes and increase accessibility.