Reflection unveils open-weight model Beam, saying it excels at coding and agentic tasks, uses 3x-4x less compute than comparable models, and nears Qwen 3.8-Max (Semafor)
Reflection says Beam is a 501B-parameter open-weight model with 23B active parameters and a full weights release due this month.
The company is positioning Beam around coding, agentic, and scientific workloads. Reflection says it was trained from scratch and is more inference-efficient than open models with similar capability. Techmeme’s item points to claims that Beam uses 3x to 4x less compute than comparable models and approaches Qwen 3.8-Max. Techmeme's note
The company is positioning Beam around coding, agentic, and scientific workloads. Reflection says it was trained from scratch and is more inference-efficient than open models with similar capability. Techmeme’s item points to claims that Beam uses 3x to 4x less compute than comparable models and approaches Qwen 3.8-Max. Techmeme's note
score 7