https://derekwang2002.github.io/ai-intelligence/en/projects/beam/ Material changes in AI models, agents, engineering and infrastructure, with primary sources and recommendations. en Mon, 02 Nov 2026 16:56:00 GMT https://derekwang2002.github.io/ai-intelligence/en/events/ev-20261005-06/ ev-20261005-06-open-weights Fri, 16 Oct 2026 15:56:07 GMT Update TRIAL <p>Beam moves from limited preview to Apache-2.0 open weights</p><p>Original story: Reflection previews Beam, a 501B-parameter open-weight coding model</p><p><strong>Why read: </strong>Beam moves from an untestable preview to Apache-2.0 open weights, changing the judgment from waiting to trial.</p><p>Recommendation: TRIAL</p><p>Sources: <a href="https://huggingface.co/blog/reflectionai/beam">Beam open release</a></p> https://derekwang2002.github.io/ai-intelligence/en/events/ev-20261005-06/ ev-20261005-06-published Tue, 06 Oct 2026 16:01:24 GMT New TRIAL <p>Beam is now available as Apache-2.0 BF16, FP8 and NVFP4 weights, with Transformers, vLLM and SGLang paths plus a hosted Hugging Face endpoint recipe. The sparse model has 501B total and 23B active parameters with 1M context. Reference hosted deployment uses eight H200 GPUs at about $40 per hour; benchmark claims remain vendor-run.</p><p>Recommendation: TRIAL</p><p>Sources: <a href="https://reflection.ai/blog/introducing-beam">Introducing Beam</a></p>