Reflection 发布开放权重模型 Beam，对标中国开源模型
原标题：Reflection debuts Beam, an open-weight AI model to rival Chinese models at lower compute cost
AI 摘要
Reflection AI 发布首个前沿开放权重模型 Beam，这是一个 5010 亿参数、230 亿激活参数的文本混合专家模型，预训练于 23.8 万亿 token，支持 100 万 token 上下文。公司声称 Beam 在高级推理基准上可媲美 Z.ai 的 GLM-5.2，并超越西方领先开放模型，同时推理算力消耗低 3-4 倍，但该性能声明尚未被独立验证。Reflection 计划本月发布模型权重和技术细节，目标客户为企业与主权国家，主打可本地训练的「AI 工厂」方案。
正文节选
Reflection AI is officially unveiling Beam, its first frontier, open-weight AI model. The two-year-old, Brooklyn-based startup claims Beam matches the performance of leading Chinese open models on advanced reasoning benchmarks at dramatically lower costs, a claim that could heat up the race to build a Western answer to DeepSeek, Qwen, and Z.ai. Reflection’s announcement confirms reporting from Axios over the weekend that the startup was close to a launch. The company shared new details in a leng