Reflection AI unveils Beam open-weight model to rival Chinese rivals
Brooklyn startup Reflection AI officially launched its new Beam model Monday with claims of matching Chinese performance at lower cost.
Model specifications and training details
The company released a text-only mixture-of-experts model containing 501 billion total parameters. This architecture uses 23 billion active parameters after being pretrained on 23.8 trillion tokens. The system features a context window of one million tokens.
Performance claims and cost efficiency
Reflection states the model scores on par with Z.ai's GLM-5.2 on advanced reasoning benchmarks. The startup asserts Beam uses three to four times less inference compute than current rivals. These performance figures have not been independently verified by outside researchers.
Target markets and positioning
The firm describes the release as a workhorse model for enterprises and developers. Reflection positions this tool against closed labs like Anthropic and OpenAI. The launch follows earlier reporting that the startup was close to releasing the software.
Reported by one outlet
Only one outlet has published this. Nothing here has been checked against a second report, so read it as that outlet's account and follow the link below for the original.
Reported by
1 independent outlet. Headline as published. Links open the original report.