Reflection's open weight model, Beam, is now accessible through Amazon SageMaker JumpStart. Beam offers strong performance in reasoning, coding, and agentic tasks, positioning itself at the forefront of its class. This integration allows customers to build advanced coding agents and automated workflows with greater control over their data and infrastructure. Beam operates within the existing SageMaker AI environment, offering scalability, audit trails, and monitoring. The model has been optimized for AWS Trainium3, providing a cost-efficient and high-performance deployment option. Reflection emphasizes that Beam provides developers with controllable frontier AI, accessible through trusted AWS infrastructure. Engineering teams can leverage Beam for tasks such as resolving software issues, generating code, and automating operations. Businesses can utilize Beam for complex, multi-step workflows and advanced scientific reasoning. Being an Apache 2.0 licensed model, Beam allows for inspection and customization, with data remaining private within the customer's AWS account. Accenture Federal Services highlights the benefits of running frontier AI securely within their own environments with full visibility. Beam is currently available in SageMaker JumpStart in US East (N. Virginia) and US East (Ohio).