Reflection unveils Beam: a 501B open-weight model built to match China's best at 3–4x less compute
Reflection AI unveiled Beam on Monday — its first frontier open-weight model, and the answer to nearly two years of speculation about what the Nvidia-backed startup had been building. The company says the 501-billion-parameter model matches the top Chinese open models on advanced reasoning benchmarks while using "3-4x less inference compute," a direct bid to become the Western answer to DeepSeek, Qwen, and Z.ai.
The announcement, published in a company blog post and reported by TechCrunch, lands two days after Axios reported Reflection was close to launch. It is the first concrete look at the startup's flagship model — and the first major American open-weight release designed from the start to compete with Chinese labs on their own terms.
The specs
Beam is a text-only mixture-of-experts model: 501 billion total parameters with 23 billion active, pre-trained on 23.8 trillion tokens, with a 1 million token context window. Reflection says it was trained on high-compute reinforcement learning to be effective at reasoning, coding, and agentic tasks "at a fraction of the token cost and inference time compute" of rivals.
For comparison, Z.ai's GLM 5.2 — the current benchmark darling of the Chinese open-weight scene — carries roughly 744 billion total parameters with 40 billion active. Reflection claims Beam scores on par with GLM-5.2 on advanced reasoning benchmarks and outperforms today's leading Western open models. Those claims have not been independently verified, and the technical paper and full details arrive later this month.
The play: AI factories

Reflection isn't selling a chatbot. It's selling institutions control. The pitch is "AI factories" — customized, locally-run AI systems built on Reflection's models and an enterprise's own proprietary data, aimed at enterprises, governments, and sovereign AI programs. Nvidia CEO Jensen Huang, whose company backs Reflection, has championed the AI-factory vision for years, and Axios reported that hedge funds and trading firms are among the first eager adopters.
The company says Beam's weights and full technical details ship later this month, with distribution through hyperscalers and neoclouds and integrations across open-source libraries at launch. A sovereign AI factory partnership is already being tested with South Korea's Shinsegae Group.
A crowded field — and a fast one
Reflection enters a market that got dramatically more competitive this year. The American open-weight bench now includes Inkling from Mira Murati's Thinking Machines Lab, released in July, and the Chinese labs keep moving: Z.ai's GLM 5.2 sets the pace Reflection measures itself against. Notably, Reflection's own benchmarks show Beam outscores Inkling on four coding tests where both report results — though Inkling is multimodal and Beam is text-only.
The company has resources to keep pace. Founded in 2024 by two former Google DeepMind researchers, Misha Laskin and Ioannis Antonoglou, Reflection has raised roughly $4.7 billion — from Nvidia, Sequoia Capital, Lightspeed Venture Partners and others, per PitchBook — at a $25 billion pre-money valuation in its last round. This summer it signed compute deals collectively worth more than $7 billion with SpaceX and Nebius to secure Nvidia GB300 chips through 2029.

What's still open
The two things that matter most — the weights themselves and the license — aren't out yet. "Open-weight" covers a wide range, from Llama-style community licenses to something far more restrictive, and enterprise buyers will read the fine print before building sovereign systems on it. The benchmark numbers are self-reported. But if Beam lands this month roughly as advertised, it changes the open-weight map: for the first time, Western buyers who want frontier-class reasoning without routing through Chinese weights or closed APIs will have a serious domestic option.
Reflection has been positioning this moment since its founding — "the global standard of intelligence will be built by someone else" if America doesn't act, as Laskin put it last year. Beam is the company's first answer. The weights are the proof.