Reflection AI has officially unveiled Beam, its first frontier open-weight model, positioning the Brooklyn-based startup as a Western answer to DeepSeek, Qwen, and Z.ai in the race for efficient open reasoning models that can run at enterprise scale without prohibitive compute costs.
Beam is a 501-billion-parameter mixture-of-experts model with 23 billion active parameters, pretrained on 23.8 trillion tokens with a 1 million token context window. The company claims it matches Z.ai's GLM-5.2 on advanced reasoning benchmarks while using "3-4x less inference compute" than rivals. Reflection calls it a "workhorse model" for enterprises, the public sector, and developers who need frontier-level reasoning without paying for closed-model API pricing. To compare, Z.ai's GLM-5.2 has roughly 744 billion total parameters with 40 billion active.
The startup was founded in 2024 by two former Google DeepMind researchers and has raised roughly $4.7 billion from backers including Nvidia, Sequoia Capital, and Lightspeed Venture Partners. Its last round valued the company at $25 billion pre-money. This summer, Reflection signed deals worth more than $7 billion with SpaceX and Nebius to secure Nvidia GB300 chip access through 2029, locking up the compute needed to train and serve frontier models.
Reflection is targeting enterprises and sovereign nations with its "AI factories" concept — letting institutions build customized local AI systems trained on their own proprietary data. Nvidia CEO Jensen Huang has championed this idea, and the vision benefits Nvidia directly since its GPUs would power those systems. The company has already begun testing a sovereign AI factory partnership with Shinsegae Group in South Korea. Beam's weights and full technical details will be released this month, with distribution through hyperscalers and neoclouds.