Poolside Introduces Open-Weight Coding Models Laguna xs 2.1 and Laguna s 2.1

LinkPoolsideTool

Poolside has launched open-weight agentic coding models. While Laguna xs 2.1 is designed for local on-device deployment, Laguna s 2.1 provides high-reasoning capabilities via desktop apps, CLI, OpenRouter, and Vercel AI Gateway.

Key points

  1. Laguna s 2.1 features 118B total parameters (8B active) and a 1M token context window.
  2. Laguna xs 2.1 includes 33B parameters (3B active) with a 256k context window and runs on-device.
  3. Access is provided through desktop tools, the 'pool' CLI (v1.0.6), OpenRouter, and Vercel AI Gateway.
  4. The 'Model Factory' engineering series shares research insights, covering Grace Blackwell C2C activation offloading and reinforcement learning from code execution.

Why it matters

Open-weight models with agentic coding capabilities allow engineering teams to maintain control over their infrastructure while utilizing frontier-class reasoning. A low active parameter count combined with large context sizes makes local deployment viable.

Analysis

By adopting a sparse mixture-of-experts design (such as 8B active out of 118B), Poolside balances reasoning power and computational costs. Their release across standard developer surfaces like CLI tools and third-party gateways lowers adoption barriers for software engineering workflows.

Mentioned links

Found in

Summary generated by AI (Gemini 3.8 Flash, 27 September 2026). It may contain errors – the original source is authoritative.

Related entries

Similar in content, found by the AI search.

  • Link:Poolside

    Laguna-S-2.1: Open-Source MoE Model by Poolside for Agentic Coding

    Poolside has released Laguna-S-2.1, a Mixture-of-Experts language model featuring 118B total parameters with approximately 8B activated parameters per token. Targeted at software engineering and long-horizon agentic workflows, the model is available under the permissive OpenMDW-1.1 license.

    AI & LLMs· Tool

  • Link:Kwaipilot

    KAT-Coder-V2.5-Dev: Open-Weight MoE Model for Coding Agents

    Kwaipilot has released KAT-Coder-V2.5-Dev, an open-weight Mixture of Experts model based on Qwen3.6-35B-A3B. Featuring 35 billion total and 3 billion activated parameters, it has been specifically post-trained for agentic software engineering tasks.

    AI & LLMs· Tool

  • Link:Anomaly

    OpenCode: Open-source AI coding agent

    OpenCode is an open-source AI coding agent running across terminals, IDEs, and desktop environments, offering integration with over 75 model providers.

    AI & LLMs· Tool

  • Link:Ollama

    Ollama: Platform for Open Models and Coding Agents

    Ollama provides infrastructure to run open models alongside coding agents. Beyond running models locally, it offers hosted cloud models with dedicated capacity and integrations for developer tools.

    AI & LLMs· Tool

  • Link:Kilo Code

    Ring-2.6-1T: Free inclusionAI Thinking Model on Kilo Code

    inclusionAI has released Ring-2.6-1T, a 1T-parameter-scale thinking model featuring 63 billion active parameters. Optimised for coding agents, the model is accessible free of charge via Kilo Code and OpenRouter.

    AI & LLMs· Tool

Back to the knowledge base

Let's talk about your project

Get in touch

Locations

  • Mattersburg
    Johann Nepomuk Bergerstraße 7/2/14
    7210 Mattersburg, Austria
  • Vienna
    Ungargasse 64-66/3/404
    1030 Wien, Austria

Parts of this content were created with the assistance of AI.