---
format: "aidr-story-markdown/v1"
id: "437725e3ef9cff60862822e562d7a15ed3ff78a28f5499d601b1513122c7c041"
canonical_url: "https://aidr.today/437725e3?lang=en"
title: "Reflection AI releases Apache 2.0-licensed weights for Beam, a 501B-parameter model, trained with RL that penalized unnecessary tokens (Matt Marshall/VentureBeat)"
lang: "en"
requested_lang: "en"
available_langs: ["en","vi"]
translation_fallback: null
fallback_fields: []
published_at: "2026-10-16T16:40:02.000Z"
category: "Models"
topics: ["open-source"]
source_urls: ["https://www.techmeme.com/261016/p15"]
summary: "Reflection AI released Apache 2.0-licensed weights for Beam, a 501B-parameter model trained with RL that penalized unnecessary tokens, positioning it against Chinese open models."
---

# Reflection AI releases Apache 2\.0\-licensed weights for Beam, a 501B\-parameter model, trained with RL that penalized unnecessary tokens \(Matt Marshall/VentureBeat\)

> [Open the canonical story](<https://aidr.today/437725e3?lang=en>)

**Published:** 2026-10-16T16:40:02.000Z
**Category:** Models
**Topics:** open\-source

## Summary

Reflection AI released Apache 2\.0\-licensed weights for Beam, a 501B\-parameter model trained with RL that penalized unnecessary tokens, positioning it against Chinese open models\.

## Sources

- [Story source](<https://www.techmeme.com/261016/p15>)