MoE articles

3 articles about MoE, newest first.

• Xiaomi MiMo v2.6: A New Frontier for Open-Weight Models — Explore Xiaomi's MiMo v2.6, a trillion-parameter open model rivaling GPT 5.6 Soul in agentic coding and scientific research, including its RL training costs.
• Kimi K3: How a 2.8T Open Model Uses MoE, Delta Attention & More to Compete with GPT‑5 — Explore Kimi K3’s 2.8‑trillion‑parameter architecture, Mixture‑of‑Experts, Stable LatentMoE, Delta Attention, and Attention Residual, and why they matter for open‑source LLMs.
• Kimi K3: The Open‑Weights Frontier for Agentic Coding — Kimi K3 is a 3‑trillion‑parameter MOE model with 1M context, native vision, and strong coding benchmarks, aiming to set a new open‑source frontier for specialized AI agents.