
Poolside's Laguna XS.2: A Free Open-Source AI Model That Runs Locally on Your Laptop
U.S. startup Poolside launched Laguna XS.2, an Apache 2.0 open-source AI coding model that runs on a single GPU without internet — challenging the dominance of proprietary models from OpenAI and Anthropic.
Laguna XS.2: Open-Source AI You Can Run Offline
Poolside, a San Francisco AI startup founded in 2023, just launched two new AI models — and the smaller one is completely free and open source. Laguna XS.2 is a 33-billion parameter Mixture of Experts model with 3 billion active parameters, released under an Apache 2.0 license.
The key selling point: you can download it and run it on your desktop or laptop without an internet connection. Completely private, completely secured, completely free.
Why This Matters for Developers
Until now, the open-source AI space has been dominated by Chinese companies like DeepSeek and Xiaomi. U.S. labs mostly competed on proprietary, paid models. Poolside breaks that pattern by offering a high-performing, locally-runnable model trained entirely from scratch in the U.S.
For developers, this means access to capable agentic coding AI without recurring API costs, without sending code to external servers, and without vendor lock-in. You can fine-tune, quantize, or serve agents on a single GPU.
The Bigger Picture: Two Models, Two Strategies
Poolside released two models:
- Laguna M.1: A proprietary 225B parameter flagship for complex enterprise and government workloads
- Laguna XS.2: The open-source 33B model for local agentic coding
Even the larger M.1 is free temporarily through Poolside's API and partners like OpenRouter, Ollama, and Baseten. This "try before you commit" approach is becoming the new standard for AI model distribution.
How to Get Started
You can download Laguna XS.2 from Hugging Face or pull it through Ollama with a single command. Poolside also launched "shimmer," a web-based, mobile-optimized coding environment, and "pool," a coding agent harness — both designed to work with Laguna models out of the box.
The model was trained using Poolside's proprietary Muon optimizer, which achieves training approximately 15% faster than standard methods. It's purpose-built for agentic workflows: writing code, using tools, and taking autonomous actions.
Common Questions (FAQ)
Q1: Can Laguna XS.2 really run on a laptop? A1: Yes. With only 3 billion active parameters in a MoE architecture, it's designed to run efficiently on consumer-grade GPUs. Quantized versions will run on even modest hardware.
Q2: How does it compare to GPT or Claude for coding? A2: It's not at the frontier level of GPT-5.5 or Claude Opus 4.7 for the most complex tasks. But for everyday coding, debugging, and agentic workflows, it offers competitive performance at zero cost.
Q3: Is it safe for enterprise use? A3: The Apache 2.0 license allows commercial use with minimal restrictions. Poolside specifically designed it for environments where data privacy and offline operation are requirements.
Stay ahead of the AI curve. Follow @AiForSuccess for daily insights.
📬 Want more AI solopreneur insights?
Subscribe to our weekly newsletter →Related Articles

AI Model API Aggregation Platforms: From Simple Proxies to Enterprise AI Hubs
AI API aggregation platforms are evolving beyond protocol translation. Discover how these platforms are becoming essential infrastructure for enterprise AI adoption.

AI Jobs Explosion: 12x Increase in AI Positions Signals Massive Talent Demand
The AI job market has exploded with a 12x increase in AI positions since 2025. Discover what's driving this talent rush and what it means for your career in artificial intelligence.

Anthropic's Claude Code Source Leak: 1900 Files, 500K Lines of Code Gone Public
In June 2026, Anthropic accidentally published nearly 1,900 source files and 500,000 lines of Claude Code's core codebase. Here's what the leak revealed and why it matters for AI developers.