Reflection's Beam becomes the most capable open-weight model built outside China
The Decoderen

Reflection has released Beam, its first open-weight model. The mixture-of-experts system activates just 23 billion of its 501 billion parameters per token and aims to match GLM 5.2 on coding and reasoning while using three to four times less compute. The strategy against Chinese open-weight rivals like Deepseek and Qwen is efficiency over raw performance. The article Reflection's Beam becomes the most capable open-weight model built outside China appeared first on The Decoder .
This is a short summary published by AI Global Wire. The full article is owned and hosted by The Decoder — open it there to read it in full.
Read the full story at The Decoder- DeepSeek
- Verktyg
Related AI news
- OpenAI agents tried to hack Wikipedia tools and flooded it with trafficArs Technica AI · October 6, 2026
- Singapore security firm V-Key takes stake in CloudsineAITech in Asia · October 6, 2026
- Researchers stretch LeCun's JEPA AI into a universal world model that works from physics to biologyThe Decoder · October 6, 2026
- SAP expands Joule into an agentic work layer as Autonomous Enterprise goes liveSiliconANGLE · October 6, 2026
- CATL and Tencent back Deepseek's ballooning funding round as the AI startup eyes a 2027 IPOThe Decoder · October 6, 2026
- Open AI börjar vattenstämpla AI-genererade texterComputer Sweden · October 6, 2026