Reflection AI releases Apache 2.0-licensed weights for Beam, a 501B-parameter model, trained with RL that penalized unnecessary tokens (Matt Marshall/VentureBeat)

Reflection AI has released the Apache 2.0-licensed weights for Beam, a 501B-parameter model. This model was trained using reinforcement learning (RL) that specifically penalized unnecessary tokens. This move positions Beam as Reflection AI's first open model, aiming to compete with existing open models, particularly those from Chinese companies.

This marks Reflection AI's first open model release, unlike other companies that have previously focused on closed-source or proprietary models.