Beam — Reflection AI's first open-weight model, introduced 5 October 2026. It is a 501B-parameter sparse Mixture-of-Experts model with 23B active parameters built for coding and agentic work, positioned against Chinese open models such as GLM-5.2 at far lower inference compute, and needs roughly 280GB of VRAM to self-host.
reflection.ai30 stories about Beam
searching across sources...Beam is a large language model by Reflection AI. Beam — Reflection AI's first open-weight model, introduced 5 October 2026. It is a 501B-parameter sparse Mixture-of-Experts model with 23B active parameters built for coding and agentic work, positioned against Chinese open models such as GLM-5.2 at far lower inference compute, and needs roughly 280GB of VRAM to self-host. This page tracks 30 recent news stories about Beam, curated from 30+ sources and updated every 15 minutes.
First tracked Oct 2026. Most-citing sources: Medium (6), TechCrunch AI (1), Marktechpost (1), AIMiracle.ai (1). Data aggregated by Best AI News Today from 30+ sources.
3 sources
by Menna Adly — The open weights sound historic until the serving math demands 501 gigabytes of VRAM.
Reflection AI opened early access to Beam, its 501B open-weight model, and plans to release the weights under Apache 2.0 later in October.
by MohamedAbdelmenem — The memory math behind 23B active routing, and the three-step audit teams need on day one.
A US lab says it matched China's best open models with less compute. What Reflection AI Beam is, who built it, and what is still unproven. The post Reflection AI Beam: 501B Open Model Takes On China...
2 sources
by YouShin kim — 1. 한눈에 보는 핵심 요약 (Executive Summary)
2 sources
Beam is Beam — Reflection AI's first open-weight model, introduced 5 October 2026. It is a 501B-parameter sparse Mixture-of-Experts model with 23B active parameters built for coding and agentic work, positioned against Chinese open models such as GLM-5.2 at far lower inference compute, and needs roughly 280GB of VRAM to self-host.
Beam is available at https://reflection.ai. It's developed by Reflection AI.
As of today, there are 30 recent stories about Beam. Recent headlines include: Reflection AI's Beam sells governments AI independence; Nvidia in talks to acquire it; Mistral Large 4 vs Kolibri vs Beam: Which Open-Weight LLM?; Reflection Previews Beam AI Coding Model, Optimized for Efficiency. Updated every 15 minutes.
Beam is part of the LLM model category. Browse all comparable AI llm models with live news mentions and benchmarks.
AI now shows up in every job description. Ten beginner AI courses compared by price, free trial and what each does best — from free no-code starters to certificates from Google and top universities.
AI beginner course
