Beam 1.0

Beam 1.0 is Reflection AI's first open-weight model: a sparse Mixture-of-Experts reasoning model with 501.3B total parameters and 23.4B active per token (52 layers: 3 dense + 49 MoE; 256 experts; gated GQA with sliding-window attention; 152K vocabulary), pretrained on 23.8T tokens and post-trained with large-scale asynchronous RL. It targets coding, long-horizon agentic work, and tool use, with a 1M-token context window, up to 128K output tokens on the Reflection API, text-only input/output, and mandatory reasoning with low/medium/high/xhigh/max effort levels. Released under Apache 2.0 in BF16, FP8, and NVFP4 with a DFlash speculative-decoding drafter; available on Hugging Face, the Reflection API (Beam-1.0-501B-A23B), OpenRouter, and other partners.