In this paper, we propose a Large Language Models (LLMs) based Vid2Seq multimodal model capable of processing videos in just one stage. With the predominant increase in the video content platforms, we have witnessed exponential increase in volume of video content and the need for dense video understanding and captioning has also grown. This allows a user to search specific video content based on their search query and at the same time helps the video content generator in producing specific content summaries and captions for the newly generated contents. The Vid2Seq model is a visual language model that generates video sequences with time tokens to predict the dense video captions including the temporal grounding. The Vid2Seq accepts multimodal inputs in the form of video and transcribed speech and then generates the video sequences and their textual descriptions as output. In the proposed model, we have utilized the promising and advanced capabilities of LLMs to generate human-like text descriptions and captions.