Reflection AI introduced Beam, a 501-billion-parameter sparse MoE open-weight model it says is strong at coding and agentic tasks, competitive with GLM-5.2, approaching Qwen 3.8-Max,…
Reflection AI introduced Beam, a 501-billion-parameter sparse MoE open-weight model it says is strong at coding and agentic tasks, competitive with GLM-5.2, approaching Qwen 3.8-Max,…