The field of large language models (LLMs) is rapidly advancing, with a focus on improving their reasoning and optimization capabilities. Recent research has explored the use of LLMs for solving complex optimization problems, including NP-hard tasks, and has proposed novel frameworks and techniques for training and evaluating these models. One notable direction is the development of methods that combine LLMs with external tools and solvers to improve their performance on mathematical reasoning tasks. Another area of research is the investigation of techniques for generating challenging problems and evaluating LLMs' ability to solve them. Overall, the field is moving towards developing more advanced and generalizable LLMs that can tackle a wide range of complex tasks. Noteworthy papers include NP-Engine, which proposes a comprehensive framework for training and evaluating LLMs on NP-hard problems, and QueST, which introduces a novel framework for generating challenging coding problems. SolverLLM is also notable for its training-free framework that leverages test-time scaling to solve diverse optimization problems.