Today for AI
HOT RADAR
llmHEAT 0.0°

AI CLUSTERED EVENT · 10/5/2026

Reflection Debuts Open-Weight Beam Model to Rival Chinese AI

3 reports archived3 independent sourcesupdated 10/6/2026, 6:28:43 AM
Synthesis & Latest Updates
3 Sources Cross-Validated

US startup Reflection AI unveiled Beam, its first frontier open-weight model, claiming it matches leading Chinese models like DeepSeek and Qwen on reasoning benchmarks at significantly lower compute costs. The 501B-parameter Mixture-of-Experts model features 23B active parameters and a 1M token context window, with weights and technical details scheduled for release this month. While performance claims await independent verification, Beam positions itself as a Western alternative in the intensifying open-source LLM race.

LATEST/US startup Reflection AI unveiled Beam, its first frontier open-weight model, claiming it matches leading Chinese models like DeepSeek and Qwen on reasoning benchmarks at significantly lower compute costs. The 501B-parameter Mixture-of-Experts model features 23B active parameters and a 1M token context window, with weights and technical details scheduled for release this month. While performance claims await independent verification, Beam positions itself as a Western alternative in the intensifying open-source LLM race.

TIMELINECoverage timeline

Total 3 reports · Latest first
  1. Latent SpaceT1·68 pts

    Reflection Launches Beam: 501B-Param US-Trained Open Model

    Original: [AINews] Reflection Beam - 501B-A23B American Open Model

    • Reflection ends stealth mode by launching its first self-developed LLM, Beam (501B-A23B).
    • The model emphasizes being 'trained from scratch in the US,' catering to market demands for geopolitical compliance.
    • Performance benchmarks indicate it does not yet surpass top-tier open models like GLM 5.3 or Kimi K3.
  2. IT之家 · 智能时代T2·78 pts

    Reflection AI Launches 501B Open-Weight Model Beam

    Original: 叫板 DeepSeek、Kimi，英伟达投资的 Reflection AI 发布旗下首款开放权重 AI 模型 Beam

    • Reflection AI launches Beam, its first open-source model backed by Nvidia
    • Uses MoE architecture: 501B total parameters, 23B active
    • Core capabilities focus on code generation and agent tasks
    • Benchmark performance approaches Qwen 3.8-Max, targeting GLM-5.2
  3. TechCrunch AIT2·78 pts

    Reflection Debuts Open-Weight Beam Model to Rival Chinese AI

    Original: Reflection debuts Beam, an open-weight AI model to rival Chinese models at lower compute cost

    • Beam is Reflection AI's first frontier open-weight model, positioned as a Western rival to DeepSeek and Qwen.
    • It utilizes a 501B total parameter Mixture-of-Experts architecture with 23B active parameters and supports a 1M token context window.
    • The company claims comparable reasoning performance to top Chinese models at a fraction of the inference cost, though these metrics remain unverified.