Reflection AI launches Beam, an open-weight model to rival China

Reflection AI is officially unveiling Beam, its first frontier, open-weight AI model. The two-year-old, Brooklyn-based startup claims Beam matches the performance of leading Chinese open models on advanced reasoning benchmarks at dramatically lower cost, a claim that could heat up the race to build a Western answer to DeepSeek, Qwen and Z.ai. The announcement confirms weekend reporting from Axios that the startup was close to a launch.

In a blog post Monday, Reflection described Beam as a text-only mixture-of-experts model trained with 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. Beam is a 501-billion-parameter model with 23 billion active parameters, pre-trained on 23.8 trillion tokens, with a 1 million token context window; by comparison, Z.ai’s GLM-5.2 has roughly 744 billion total parameters with 40 billion active.

Reflection’s performance claims haven’t been independently verified, but the company says that on advanced reasoning benchmarks, Beam scores on par with Z.ai’s GLM-5.2 and outperforms today’s leading Western open models while using three to four times less inference compute. Reflection calls it a “workhorse model” for enterprises, the public sector and developers, positioning itself against closed labs such as Anthropic and OpenAI, against popular Chinese open models, and against Western players including Mistral, Meta and Cohere. Reflection’s most direct U.S. rival may be Inkling, the open model from Mira Murati’s Thinking Machines Lab released in July; Reflection’s own benchmarks show Beam outscoring Inkling on four coding tests where both report results, though Inkling is multimodal while Beam is text-only.

Founded in 2024 by two former Google DeepMind researchers, Reflection has raised roughly $4.7 billion from backers including Nvidia, Sequoia Capital and Lightspeed Venture Partners, according to PitchBook, with its last round valuing the company at a $25 billion pre-money valuation. The startup has also locked up compute, signing deals worth more than $7 billion this summer with SpaceX and Nebius to secure access to Nvidia’s GB300 chips through 2029. Reflection is aiming Beam and future models at enterprises and sovereign nations through a pitch it calls “AI factories” — letting institutions train Reflection’s models on their own proprietary data — an idea Nvidia CEO Jensen Huang has long championed and that would also benefit Nvidia’s chip sales. Reflection has begun testing the concept with South Korea’s Shinsegae Group, and says it will release Beam’s weights and technical details this month, with distribution through hyperscalers and neoclouds at launch. Reflection did not respond to TechCrunch’s request for comment.

Why it matters

If Reflection’s efficiency claims hold up under independent testing, it would strengthen the case that non-Chinese open models can match frontier Chinese labs on cost as well as capability — a factor that matters directly for enterprises and governments picking which open model to build on.