The recent introduction of the LlamaV-o1 model has garnered attention for its capabilities in visual reasoning within large language models (LLMs). This model reportedly outperforms other competitors, including Gemini-1.5-flash, GPT-4o-mini, Llama-3.2-Vision-Instruct, Mulberry, and Llava-CoT. Key features of LlamaV-o1 include a new benchmark for step-by-step visual reasoning, an innovative evaluation metric, and a curriculum learning approach designed to enhance accuracy and speed. Additionally, LlamaIndex has launched its first publicly released model, vdr-2b-multi-v1, which facilitates multimodal search by integrating text and image processing. Google has also made strides in this area with the rollout of its Vertex AI RAG Engine, aimed at enabling scalable multimodal search capabilities.