The field of robotic manipulation is witnessing significant advancements with the integration of vision-language models (VLMs). Recent developments focus on enhancing the scene context awareness and semantic understanding of VLMs, enabling them to better generalize to unseen tasks and environments. Researchers are exploring innovative approaches to improve the performance of VLMs in robotic manipulation, including the incorporation of chain-of-thought reasoning, depth perception, and task-oriented region of interest detection. These advancements have led to improved success rates in various benchmarks and real-world scenarios, demonstrating the potential of VLMs in achieving robust and autonomous robotic grasping and manipulation. Noteworthy papers in this area include: 3D CAVLA, which achieves an average success rate of 98.1% and an 8.8% absolute improvement on unseen tasks. UniDiffGrasp, which enables precise and coordinated open-vocabulary grasping in complex real-world scenarios with grasp success rates of 0.876 in single-arm and 0.767 in dual-arm scenarios. Through the Looking Glass, which assesses image common sense consistency using Large Vision-Language Models and achieves state-of-the-art performance on the WHOOPS and WEIRD datasets. Training Strategies for Efficient Embodied Reasoning, which provides a better understanding of why chain-of-thought reasoning helps vision-language-action models and introduces two simple and lightweight alternative recipes for robot reasoning. ORACLE-Grasp, which leverages Large Multimodal Models as semantic oracles to guide grasp selection without requiring additional training or human input. From Seeing to Doing, which proposes a novel vision-language model that generates intermediate representations through spatial relationship reasoning, providing fine-grained guidance for robotic manipulation. ManipBench, which evaluates the low-level robot manipulation reasoning capabilities of VLMs across various dimensions. Unfettered Forceful Skill Acquisition, which demonstrates that eliciting wrenches allows VLMs to explicitly reason about forces and leads to zero-shot generalization in a series of manipulation tasks. PointArena, which probes multimodal grounding through language-guided pointing and evaluates multimodal pointing across diverse reasoning scenarios.