Developed by Twelve Labs, Pegasus-1 is a powerful video-to-text generation model capable of generating video titles, summaries, and custom text outputs. With 80 billion parameters, Pegasus-1 outperforms previous video-language models by 61% on the MSR-VTT dataset and 47% on the Video Descriptions dataset. Users can leverage Pegasus-1's API to generate various text outputs for videos, including titles, summaries, chapters, and custom formats. Pegasus-1 comprehensively considers visual, audio, and speech information from the video, resulting in more comprehensive and accurate text generation compared to existing solutions.