{"slug": "moe-analysis-qwen-3-5-3-6-35b-a3b", "title": "MoE Analysis Qwen[3.5|3.6]35B-A3B", "summary": "An analysis of two Mixture-of-Experts models, Qwen 3.5 35B-A3B and Qwen 3.6 35B-A3B, captured per-token router statistics across every MoE layer using first-turn MT-Bench prompts, recording which experts the router selected and how confident each choice was. The study, published by G.S. Laller, reports expert-selection and router-confidence data by model, color and layer.", "body_md": "[/MoE routing](./)\n\n**G.S. LALLER**/Notes\n# MoE Analysis, which experts fire, and how sure is the router?\n\nWe took the MT-Bench prompts (first turn only), ran inference on each of them with two Mixture-of-Experts models, and captured the statistics of every MoE layer: for each token, which experts the router picked, and how confident that choice was.\n\nModel \n\nColour by \n\nLayer \n\nLoading `data/manifest.json`…", "url": "https://wpnews.pro/news/moe-analysis-qwen-3-5-3-6-35b-a3b", "canonical_source": "https://gslaller.github.io/moe_analysis.html", "published_at": "2026-09-28 23:21:51+00:00", "updated_at": "2026-09-28 23:47:44.665046+00:00", "lang": "en", "topics": ["artificial-intelligence", "machine-learning", "large-language-models", "ai-research"], "entities": ["Qwen 3.5 35B-A3B", "Qwen 3.6 35B-A3B", "G.S. Laller", "MT-Bench"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/moe-analysis-qwen-3-5-3-6-35b-a3b", "markdown": "https://wpnews.pro/news/moe-analysis-qwen-3-5-3-6-35b-a3b.md", "text": "https://wpnews.pro/news/moe-analysis-qwen-3-5-3-6-35b-a3b.txt", "jsonld": "https://wpnews.pro/news/moe-analysis-qwen-3-5-3-6-35b-a3b.jsonld"}}