Nvidia-Backed Reflection Unveils Open AI Model, Taking on China

Reflection AI, an artificial intelligence startup from two former Google DeepMind researchers, has unveiled a new open-weight model that it says rivals leading options in the US and China.

The model, called Beam, is designed to perform well for business customers and developers on coding and AI agent tasks while costing less than many competitors, the company said on Monday. It said Beam scores comparably on advanced reasoning benchmarks to GLM-5.2, a flagship open model from China’s Z.AI.

Reflection intends to release more technical details about the model later this month.

While China is still widely viewed as behind the US in AI, Chinese developers like DeepSeek and Moonshot AI have gained ground by offering cheaper and customizable open model. Top US firms, including OpenAI and Anthropic PBC, have primarily focused on closed or proprietary offerings as they work to build up their enterprise businesses.

Reflection emerged from stealth last year with a focus on developing open AI models. The firm is backed by Nvidia Corp. and other investors. It has held talks to raise $2.5 billion at a valuation of $25 billion, the Wall Street Journal previously reported.

The startup has also struck billion-dollar deals with Elon Musk’s SpaceX and cloud provider Nebius Group NV to secure computing power to develop its models.

The model release stands to boost the American government’s efforts to export the US and its allies’ AI abroad. Earlier this year, Reflection was part of a US-brokered deal to build sovereign AI in Korea. Officials are planning to cobble together more of these deals promoting exports of US tech, which are designed to prevent the spread of Chinese alternatives.