The field of black-box optimization is moving towards more efficient and robust methods, with a focus on leveraging structural information and transfer learning. Recent work has explored the use of pre-trained models, counterfactual inference, and robust Bayesian optimization to improve sample efficiency and handle uncertainty. Notably, the development of community platforms and open-source tools is facilitating collaboration and accelerating progress in the field. Some noteworthy papers include: OptunaHub, which introduces a community platform for centralizing black-box optimization methods and benchmarks. ZeroShotOpt, which presents a pre-trained model for continuous black-box optimization tasks that achieves robust zero-shot generalization on unseen benchmarks. BONSAI, which proposes a robust Bayesian optimization framework that leverages partial structural knowledge to improve sample efficiency and scalability. Counterfactual Credit Guided Bayesian Optimization, which introduces a novel framework that quantifies the contribution of individual historical observations to accelerate convergence to the global optimum.