Latent Space·· 10 d ago
[AINews] Reflection Beam - 501B-A23B American Open Model
[AINews] Reflection Beam - 501B-A23B American Open Model
AI summary
Reflection announced Beam, a text-only 501B-total / 23B-active MoE for coding, agentic and scientific work. It was trained from scratch, and full weights under Apache 2.0 are due this month.
Selection record
Threshold 76Media and individualsFirst 74Second 78
AdmittedSum of both 152 ≥ twice the threshold 152
- Source tier
- Media and individuals; this tier's threshold is 76
- Pre-filter
- passed：发布Beam开源大模型及AI行业动态
- Why it was chosen
- Reflection emerges from stealth with a 501B-A23B open MoE, plus training scale and SWE-bench results, offering a point of comparison with the current US and Chinese open-model landscape.
- Same event
A model scores each item twice, independently, against one written standard, out of 100. An item is admitted only when the two scores add up to twice the threshold. The threshold is set per source tier.
Source: Latent Space · latent.space