Unverified50% confidenceFactExact time
Reflection is building a mixture-of-experts (MoE) model called Beam with 501B parameters and plans to release it as open weights.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
11/5/2026
First Seen
Valid until: 2/3/2027
Sources
Related Entities
Related Claims
UnverifiedThe Beam model is positioned to let teams run and customize it on their own hardware, infrastructure, or in the cloud.63% similarUnverifiedMeta released the Muse Spark Open open-source large model, which uses a MoE architecture, has a total parameter count of 816B and 42B activated parameters, supports a 1M token context, and is released in the form of open weights.60% similarUnverifiedThe SkyRL team announced Day-0 reinforcement learning support for Reflection's Beam model, meaning the Beam model can be plugged into the full RL training pipeline on the day of its release.58% similarUnverifiedMiMo V2.6 Pro uses a Mixture of Experts (MoE) architecture, and the Flash version is likewise a MoE architecture.55% similarUnverifiedMeta released an open-weight model named Muse Glimmer52% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1137593API
curl https://kongchang.com/api/v1/knowledge/claims/1137593MCP
get_claim(id=1137593)