Reflection AI just dropped its Beam model with a massive 501B open‑weight Mixture‑of‑Experts. It’s crushing coding tasks and showing real agentic work. Curious how sparse MoE can boost dev productivity? Dive in! #ReflectionAI #BeamModel #SparseMoE

🔗 aidailypost.com/news/reflect...

LLaDA-MoE, a new diffusion language model, has 7 B total parameters but only activates about 1.4 B during inference after training on roughly 20 trillion tokens. Read more: https://getnews.me/llada-moe-launches-sparse-moe-diffusion-language-model/ #lladamoe #diffusion #sparsemoe

🚀 Cohere just dropped Command A+ 218B, a sparse MoE model that runs on just two H100 GPUs with QAD. Open‑source power meets massive scale—curious how they did it? Dive in! #Cohere #SparseMoE #H100

🔗 aidailypost.com/news/cohere-...

Weipu Chen, Zhuangzhuang He, Fei Liu
When SparseMoE Meets Noisy Interactions: An Ensemble View on Denoising Recommendation
https://arxiv.org/abs/2409.12730

Kepeng Lin, Qizhe Zhang, Rui Wang, Xuehai Hu, Wei Xu: PlantBiMoE: A Bidirectional Foundation Model with SparseMoE for Plant Genomes https://arxiv.org/abs/2512.07113 https://arxiv.org/pdf/2512.07113 https://arxiv.org/html/2512.07113