Reflection AI has unveiled Beam, its first open-weight model, and is pitching it as a lower-cost competitor to leading Chinese systems.
The New York City- and Brooklyn-based startup said Beam matches the performance of leading Chinese open models on advanced reasoning benchmarks, while running at 'a fraction of the token cost and inference time compute' of rivals. The company also said Beam is '3-4x less inference compute' than leading Western open models.
TechCrunch reported that Beam is a 501-billion-parameter model with 23 billion active parameters. It was pretrained on 23.8 trillion tokens and has a 1 million token context window. Reflection says the model is text-only and built as a mixture-of-experts system trained with high-compute reinforcement learning for reasoning, coding, and agentic tasks.
The company is positioning Beam as a 'workhorse model' for enterprises, the public sector, and developers. It is also aiming Beam and future models at enterprises and sovereign nations through what it calls 'AI factories,' a product that would let institutions build customized, local AI systems using their own proprietary data.
Reflection said Beam will be released with its weights and full technical details this month, with distribution through hyperscalers and neoclouds and integrations across open-source libraries at launch.
The company is entering a crowded field. Fortune reported that China is currently dominating open-source AI with models such as Kimi K3 from Moonshot and DeepSeek's V4. TechCrunch said Reflection is also positioning itself against Anthropic, OpenAI, Mistral, Meta, and Cohere, while Nvidia's Nemotron is the top-ranked U.S.-based model on OpenRouter's leaderboard.
Reflection was founded in 2024 by former Google DeepMind researchers Misha Laskin and Ioannis Antonoglou. The company has raised roughly $4.7 billion from backers including Nvidia, Sequoia Capital, and Lightspeed Venture Partners, and it was last valued at $25 billion pre-money.
It has also locked up major compute. This summer, Reflection signed deals collectively worth more than $7 billion with SpaceX and Nebius to secure access to Nvidia's GB300 chips through 2029.
The numbers come first. Beam is not being sold as a slogan; it is being sold as a cost curve. If Reflection's claims hold, the market has already voted for the next battleground in AI: not just model quality, but efficiency, control, and who can deploy capital without wasting it.
That matters for the taxpayer, for private enterprise, and for national sovereignty. Open systems give institutions more control over data and costs, and that is exactly where the competitive edge may be won or lost.
