Reflection has introduced Beam, its first open-weight model with 501 billion total parameters and 23 billion active per token.
Reflection has introduced Beam, its first open-weight model with 501 billion total parameters and 23 billion active per token.
The MoE model is designed for coding, reasoning and agentic workloads and was pretrained on 23.8 trillion tokens. Reflection plans to release the weights under the Apache 2.0 license later this month, alongside its technical report, model card and developer tools.