Reflection AI 发布开源权重智能体编程模型 Beam，可在 AWS 部署

AI 摘要

Reflection AI 发布开源权重模型 Beam，面向智能体编程和长周期任务，支持代码维护、工具调用和多步工作流。Beam 为稀疏混合专家模型，总参数 5010 亿、激活参数 230 亿，有效上下文长度 100 万 token，以 Apache 2.0 许可在 Hugging Face 发布 BF16、FP8、NVFP4 检查点。AWS 博客介绍了在 SageMaker JumpStart 实时端点或 HyperPod Inference 上部署 Beam 的路径，并说明参数规模对显存、分片和并行策略的影响。

正文节选

Introducing Reflection Beam: an open-weight model for agentic coding on AWS Reflection AI introduced Beam, an open-weight model for agentic coding and long-horizon tasks such as maintaining code, using tools, interpreting results, and completing multi-step workflows. The open-weight release gives developers and researchers control over the runtime, deployment environment, and customization process. In this post, we explain Reflection’s vision and show how you can deploy Beam on AWS. Start with a

发布时间：2026-10-16 23:23

抓取时间：2026-10-17 00:22

来源机构：AWS