{
  "slug": "jetbrains-mellum2-thinking-12b-a2-5b",
  "order": 77,
  "name": "JetBrains Mellum2 Thinking (12B-A2.5B)",
  "baseTitle": "JetBrains Mellum2 Thinking 12B-A2.5B",
  "company": "JetBrains",
  "country": "Czech Republic",
  "flag": "🇨🇿",
  "date": "2026-06-01",
  "decoderType": "Sparse MoE",
  "attention": "GQA with 3:1 sliding-window/full attention",
  "layerMix": "21 sliding-window + 7 full GQA",
  "scale": "12B total, 2.5B active (20.8% active)",
  "totalParams": 12000000000,
  "activeParams": 2500000000,
  "context": "131,072",
  "contextTokens": 131072,
  "license": "Apache License 2.0",
  "kvCacheText": "56 KiB",
  "kvCacheKiB": 56,
  "kvCacheLevel": "Low",
  "summary": "JetBrains' Mellum2 Thinking is a compact reasoning-tuned sparse MoE for code and agentic workflows.",
  "keyDetail": "Uses 64 experts with 8 active per token, a 1024-token sliding window, and YaRN scaling on full-attention layers.",
  "image": {
    "full": "https://sebastianraschka.com/llm-architecture-gallery/images/architectures/mellum2.webp",
    "thumb": "https://sebastianraschka.com/llm-architecture-gallery/images/architectures/thumbnails/mellum2.webp",
    "alt": "Figure 1: JetBrains Mellum2 Thinking 12B-A2.5B",
    "width": 960,
    "height": 925
  },
  "articleUrl": null,
  "links": [
    {
      "label": "config.json",
      "href": "https://huggingface.co/JetBrains/Mellum2-12B-A2.5B-Thinking/blob/main/config.json",
      "title": "JetBrains/Mellum2-12B-A2.5B-Thinking"
    },
    {
      "label": "Tech report",
      "href": "https://arxiv.org/abs/2605.31268",
      "title": null
    }
  ],
  "concepts": [
    {
      "slug": "gqa",
      "label": "GQA",
      "name": "Grouped-Query Attention",
      "url": "https://llm-atlas-7or.pages.dev/concepts/gqa/"
    },
    {
      "slug": "swa",
      "label": "SWA",
      "name": "Sliding Window Attention",
      "url": "https://llm-atlas-7or.pages.dev/concepts/swa/"
    },
    {
      "slug": "moe",
      "label": "MoE",
      "name": "Mixture of Experts",
      "url": "https://llm-atlas-7or.pages.dev/concepts/moe/"
    }
  ],
  "aai": null,
  "sourceUrl": "https://sebastianraschka.com/llm-architecture-gallery/#card-jetbrains-mellum2-thinking-12b-a2-5b",
  "family": "moe",
  "url": "https://llm-atlas-7or.pages.dev/models/jetbrains-mellum2-thinking-12b-a2-5b/",
  "markdownUrl": "https://llm-atlas-7or.pages.dev/models/jetbrains-mellum2-thinking-12b-a2-5b.md"
}