Reflection

Reflection develops open-weight AI models, open-source software for building AI stacks, and infrastructure for serving or deploying models. Its first open-weight model, Beam, is described on the site as a 501B MoE model built for coding, reasoning and agentic workloads.

Reflection serves open models through its API platform or lets customers deploy them in their own environment, from private cloud to on-prem, air-gapped and edge infrastructure. Its stated customers are enterprises, governments and sovereign entities, supported by a forward-deployed team. The site cites validation on Dell AI Factory with NVIDIA, provision of its models to the U.S. Department of Energy's Genesis Mission, and a planned sovereign AI cloud in the Republic of Korea described as a 250MW data center.

The company lists offices or hiring locations in New York, San Francisco, London and Washington, D.C. Its careers page shows about 40 open roles across engineering, research, security, finance, legal and commercial functions, with some roles listed in Seoul. The founding year and headquarters are not stated on the pages reviewed.

Product: Open-weight AI models (Beam), open-source AI software and tooling, and API or self-hosted deployment of models