Beam model by Reflection is now available on Amazon SageMaker JumpStart and AWS Trainium
The first open-weight model from Reflection, Beam, is now available in Amazon SageMaker JumpStart and optimized for AWS Trainium3, offering high-performance, cost-efficient deployment options.
Reflection, a us-based company, has made beam, its first open-weight model, available on amazon sagemaker jumpstart, the model hub for amazon sagemaker ai. Beam delivers strong performance in inference, coding, and agent benchmarks as a state-of-the-art open-weight model in its class. Customers can build advanced coding agents and automated workflows while managing their own data and infrastructure and tuning the model to fit their business. Beam can be used with other models already in sagemaker ai and scales to production demands. Ported and optimized for aws trainium3, beam offers a high-performance, cost-effective option. Released under the apache 2.0 license, beam allows customers to inspect and customize the model and run it in their own aws accounts. Data remains private and is protected by aws security and compliance features that customers already trust.
Why it matters
Amazon SageMaker JumpStart is a platform that allows developers to easily discover and use AI models, while AWS Trainium provides high-performance AI infrastructure. This update enables developers to securely utilize and customize a cutting-edge open-weight model.
Regions: N. Virginia, Ohio