Reflection AI 发布首个开源模型 Beam，501B 参数采用 Apache 2.0 许可
Reflection AI 放出了 501B 的开源 MoE 模型 Beam，权重直接可用，还附了 BF16、FP8、NVFP4 三种精度版本，感兴趣的可以下载试试。
Reflection AI 推出其首个开源权重模型 Beam，总参数量 501B 的 MoE 架构。模型在 NVIDIA GB300 NVL72 上训练，在 Blackwell Ultra 上单卡开箱即达每秒 70K tokens 以上的吞吐。官方同时提供 BF16、FP8 和 NVFP4 三种权重版本以及 DFlash2 drafter。权重、tokenizer 和参考实现已在 Hugging Face 上以 Apache 2.0 许可发布，还有 40 多个合作伙伴参与生态共建，训练所用的开源工具链 Mirror CLI 一并开源。
Welcome Beam! Congrats to @ReflectionAI on launching their first open-weight model. The 501B MoE model was trained on NVIDIA GB300 NVL72 and delivers over 70K tokens/sec/GPU on Blackwell Ultra out of the box on day 0. 🤗 Beam 1.0 NVFP4 checkpoint on @huggingface : huggingface.co/reflection/Bea… L Reflection @reflection_ai Beam is now available across platforms, with open weights under the Apache 2.0 license - in collaboration with an ecosystem of 40+ partners. We're releasing: - Weights, tokenizer, and reference implementation - BF16, FP8 and NVFP4 weights and a DFlash2 drafter - The open source harness we built and trained Beam on: Mirror CLI Weights and model card on Hugging Face: huggingface.co/reflection 🔗 View Quoted Tweet 💬 5 🔄 2 ❤️ 18 👀 1362 📊 5 ⚡
