Open-Weight LineageReflection
Reflection launches Beam, an open-weight model
Reflection launched Beam, an open-weight model — a sparse mixture-of-experts architecture with 501 billion total and 23 billion active parameters per token, released under Apache 2.0 with a context of 1 million tokens
Why it matters
It is the first model from a new US lab. Company-reported scores were SWEBench Verified 81.6%, GPQA Diamond 90.6%, AIME 2026 97.7% and Terminal Bench v2.1 80.1% (on BF16), and it ships in three versions, BF16, FP8 and NVFP4, with the NVFP4 checkpoint at about 280GB
Source — Primary sources first; secondary reports are labelled “Press”
Officialhuggingface.cohttps://huggingface.co/blog/reflection-beamAdded in Week 43, 2026 · See that week