{
  "slug": "laguna-xs-2",
  "order": 62,
  "name": "Laguna XS.2 (33B)",
  "baseTitle": "Laguna XS.2",
  "company": "Poolside",
  "country": "United States",
  "flag": "🇺🇸",
  "date": "2026-04-28",
  "decoderType": "Sparse MoE",
  "attention": "Gated GQA with QK-Norm and 3:1 sliding-window/global attention",
  "layerMix": "30 sliding-window + 10 global",
  "scale": "33B total, 3B active (9.1% active)",
  "totalParams": 33000000000,
  "activeParams": 3000000000,
  "context": "131,072",
  "contextTokens": 131072,
  "license": "Apache License 2.0",
  "kvCacheText": "160 KiB",
  "kvCacheKiB": 160,
  "kvCacheLevel": "Moderate",
  "summary": "Poolside's open-weight agentic coding MoE uses mixed sliding-window/global attention and activates only 3B of 33B parameters per token.",
  "keyDetail": "Uses per-layer query-head counts, a 512-token SWA window, sigmoid MoE routing, and one shared expert alongside the top-8 routed experts.",
  "image": {
    "full": "https://sebastianraschka.com/llm-architecture-gallery/images/architectures/laguna-xs2.webp",
    "thumb": "https://sebastianraschka.com/llm-architecture-gallery/images/architectures/thumbnails/laguna-xs2.webp",
    "alt": "Figure 9: Laguna XS.2",
    "width": 960,
    "height": 913
  },
  "articleUrl": "https://magazine.sebastianraschka.com/p/recent-developments-in-llm-architectures#%C2%A73-layer-wise-attention-budgeting-laguna-xs2",
  "links": [
    {
      "label": "View in article",
      "href": "https://magazine.sebastianraschka.com/p/recent-developments-in-llm-architectures#%C2%A73-layer-wise-attention-budgeting-laguna-xs2",
      "title": null
    },
    {
      "label": "config.json",
      "href": "https://huggingface.co/poolside/Laguna-XS.2/blob/main/config.json",
      "title": "poolside/Laguna-XS.2"
    },
    {
      "label": "License",
      "href": "https://huggingface.co/poolside/Laguna-XS.2/blob/main/LICENSE.md",
      "title": null
    },
    {
      "label": "Tech report",
      "href": "https://poolside.ai/assets/laguna/laguna-m1-xs2-technical-report.pdf",
      "title": null
    }
  ],
  "concepts": [
    {
      "slug": "gqa",
      "label": "GQA",
      "name": "Grouped-Query Attention",
      "url": "https://llm-atlas-7or.pages.dev/concepts/gqa/"
    },
    {
      "slug": "qk-norm",
      "label": "QK-Norm",
      "name": "QK-Norm",
      "url": "https://llm-atlas-7or.pages.dev/concepts/qk-norm/"
    },
    {
      "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/"
    },
    {
      "slug": "attention-budgeting",
      "label": "Attention Budgeting",
      "name": "Layer-wise attention budgeting",
      "url": "https://llm-atlas-7or.pages.dev/concepts/attention-budgeting/"
    }
  ],
  "aai": null,
  "sourceUrl": "https://sebastianraschka.com/llm-architecture-gallery/#card-laguna-xs-2",
  "family": "moe",
  "url": "https://llm-atlas-7or.pages.dev/models/laguna-xs-2/",
  "markdownUrl": "https://llm-atlas-7or.pages.dev/models/laguna-xs-2.md"
}