Pegasus 1.5 from TwelveLabs is now available in Amazon Bedrock
aws ·
TwelveLabs has made its Pegasus 1.5 video language model available on Amazon Bedrock, enabling developers to convert long-form video into structured, timestamped metadata. This update allows the model to analyze videos up to two hours in length without requiring pre-indexing or ingestion pipelines. Users can process content directly from Amazon S3 URIs or base64 strings while keeping data within AWS boundaries. Key new features include time-based metadata generation, multimodal prompting with reference images, and on-the-fly processing capabilities. The model supports general text analysis as well as specific segment definitions that return JSON outputs with start and end times. It is currently available in the US West (Oregon) region and billed through AWS Marketplace. This integration aims to help organizations efficiently catalog and search large volumes of unstructured video data.
aws