Amazon has launched Amazon Titan Text Premier, a new large language model (LLM) as part of its Titan Text models. The model is now available in Amazon Bedrock, a managed service that provides a selection of high-performing foundation models from leading AI companies. The new model is intended for large-scale enterprise-grade text generation applications.
Amazon Titan Text Premier is efficient, advanced, and cost-effective, offering improved performance for Retrieval Augmented Generation (RAG) and agents. It supports a wide array of text-related tasks such as summarization, generation, classification, question-answering, and information extraction. The model adheres to responsible AI practices promoting safety, security, and trustworthiness.
This release post features two sample applications that use the new Amazon Titan Text Premier model: a Document Explorer and a chat assistant. The Document Explorer application can help users understand how to build generative AI applications on AWS. This application includes a data ingestion pipeline, document summarization, and question answering capabilities. To deploy the application, users are guided to deploy all required infrastructure, upload a PDF document to the input Amazon S3 bucket, and then interact with the document browser and Q&A features.
The second sample application is the Amazon Bedrock Agent and Custom Knowledge Base, a chat assistant which can answer questions about literature using RAG, retrieving data from a selection of Project Gutenberg books stored in an S3 bucket. A new agent is deployed by updating a file to specify the model to be used when creating the agent.
Amazon Titan Text Premier is now available in the US East (N. Virginia) Region and the availability of custom fine-tuning for this model is on preview in the same region. Further information about this model is available on the Amazon Titan product page. Also, additional details on available constructs and further documentation can be found in the AWS Generative AI CDK Constructs GitHub repository. To avoid unexpected charges, users are advised to delete the AWS CloudFormation stacks and remove all data from the S3 buckets after use.
