Hugging Face and US lab Reflection have announced the release of Beam, an open-source mixture-of-experts model featuring 501 billion total parameters and 23 billion active parameters per token. Released under the Apache 2.0 license, Beam is explicitly focused on coding and agentic use cases, supporting a one-million-token context window for long-context reasoning. The model architecture features 52 decoder layers utilizing a hybrid attention setup, combining 43 sliding-window attention layers with 9 global attention layers, alongside a deep-expert routing scheme. Beam arrives with day-0 support across standard ecosystem tools including transformers, vLLM, and SGLang, along with dedicated inference, supervised fine-tuning, and reinforcement learning recipes. Deployment options include multiple precision checkpoints such as BF16, FP8, and NVFP4, which accommodate hardware ranging from NVIDIA Hopper and Ampere chips to Blackwell GPUs. Additionally, Reflection has launched a $10,000 Hub credit program to incentivize developers building and sharing projects utilizing the new model on the Hugging Face Hub.
Welcome Beam: a 501B open model you can own, customize, and scale
Reflection has released Beam, its first open mixture-of-experts model, on Hugging Face. Featuring 501 billion total parameters and 23 billion active parameters per token, Beam is designed for coding and agentic tasks with a one-million-token context window.
Why it matters
The release provides the AI community with a massive open-source frontier-scale model optimized for long-context reasoning and agentic workflows, complete with day-0 ecosystem support and deployment recipes.