Reflection’s Beam trails top open models on coding tests but claims lower inference compute

Reflection AI has built Beam, a 501-billion-parameter open-weight model for coding and agent tasks, and plans to publish the weights under an Apache 2.0 license later this month. Beam uses a mixture-of-experts design in which 23 billion parameters fire per token, which Reflection AI says lowers inference compute, though the model trails the top open models on coding tests.

Reflection AI has built Beam, a 501-billion-parameter open-weight model for coding and agent tasks, and plans to publish the weights under an Apache 2.0 license later this month. Open-weight means developers can download the trained model and run it on their own hardware. Beam uses a mixture-of-experts design, so 23 billion of its parameters fire for any given token, which keeps the cost of each answer down. Beam does not lead the open field on … More https://www.helpnetsecurity.com/2026/10/06/reflections-beam-trails-top-open-models-on-coding-tests-but-claims-lower-inference-compute/ The post Reflection’s Beam trails top open models on coding tests but claims lower inference compute https://www.helpnetsecurity.com/2026/10/06/reflections-beam-trails-top-open-models-on-coding-tests-but-claims-lower-inference-compute/ appeared first on Help Net Security https://www.helpnetsecurity.com .