Reflection unveils Beam, a 501-billion-parameter open-weight model built for coding and agents
AI startup Reflection introduced Beam on October 5, its first open-weight model. It's a text-only mixture-of-experts model with 501 billion total parameters, of which only 23 billion are active for each word it processes, trained heavily on coding, reasoning and agent-style tasks. Reflection says Beam matches GLM 5.2 on advanced reasoning tests while using three to four times less computing power to answer, though larger open models like Kimi K3 are still ahead on raw capability. It's in early access for now, and Reflection plans to release the full weights under the Apache 2.0 licence later this month.
What it means for manufacturers and fabricators
The race among open models is now about doing the same work for less computing power, which keeps pushing down the cost of running a capable AI agent on your own quotes, drawings and paperwork.