Nvidia-backed startup Reflection has launched its first open-weight model for coding, reasoning, and agentic tasks.

The U.S. startup announced the launch of Beam on its blog Monday, comparing it to Chinese models such as Z.ai's GLM 5.2 and Alibaba's Qwen 3.8 Max, while flagging that it has the upper hand in inference compute efficiency.

Beam has been pretrained on 23.8 trillion tokens and produces results on par with or better than its competitors, Reflection said.

"These results translate into more intelligence per token, delivering strong model capabilities at lower cost, making Beam a powerful workhorse model for enterprise coding and agentic workloads," Reflection said in its blog.