Reflection AI unveils Beam, an open-weight model it says nears Qwen on coding at lower cost

Reflection AI introduced Beam, its first open-weight model: 501 billion parameters in total, 23 billion active at a time, built for coding, reasoning and agent work. Reflection says it is competitive with larger open models such as GLM 5.2 and uses three to four times less computing to answer. It is in final safety testing, with weights, a model card and a technical report due later this month.

• If you pay per token for coding or agent work, note Beam's release date and compare its price with your current model once hosts list it.
• If customer data must stay in-house, ask your IT provider whether an open-weight model like this could run on servers you control.

Source: Reflection AI. This is our short summary of their reporting. The paragraphs marked "Our read" and the things to do are SeedFoundry's opinion. The summary is drafted with AI tools. Image: Reflection AI, from its own announcement. How we source and credit