LogoTopAIHubs
icon of Laguna by Poolside

Laguna by Poolside

Foundation model company building software agents and artificial general intelligence.

Introduction

What is Laguna by Poolside

Laguna is a family of open-weight agentic coding foundation models developed by Poolside, a company focused on building artificial general intelligence. The models are designed to assist with software development tasks by understanding and generating code. Currently, two variants are available: Laguna XS.2 and Laguna M.1.

How to use Laguna by Poolside
  1. Download the CLI tool: Use the command $ pool to get started with the local agent.
  2. Select a model: Use ctrl+m to choose between Laguna XS.2 (lightweight, runs on-device) or Laguna M.1 (most capable, available via API).
  3. Interact via terminal: The pool CLI provides a terminal interface with keyboard shortcuts for efficient workflow.
  4. Use on OpenRouter: Alternatively, access Laguna models through the OpenRouter platform.
Features of Laguna by Poolside
  • Open-weight models: Both Laguna XS.2 and M.1 are released with open weights.
  • Agentic coding: Models are designed to act as coding agents, capable of understanding and executing complex programming tasks.
  • On-device capability: Laguna XS.2 (33B params, 3B active) is small enough to run locally on a device.
  • Scalable performance: Laguna M.1 (225B params, 23B active) offers long-horizon reasoning and is the largest model in the family.
  • CLI tool: The pool command-line interface provides a terminal-based environment for interacting with the models.
  • Keyboard shortcuts: Efficient navigation with shortcuts like shift+tab to cycle mode, ctrl+t to toggle tool grouping, and ctrl+g to open prompts in an editor.
Use Cases of Laguna by Poolside
  • Software development: Automate coding tasks, generate code snippets, and debug programs.
  • On-device AI: Run lightweight models locally for privacy-sensitive or offline applications.
  • Agentic workflows: Build autonomous coding agents that can plan and execute multi-step software engineering tasks.
  • Research and experimentation: Use open-weight models for fine-tuning, customization, or studying agentic AI.
FAQ

Q: What does "open-weight" mean? A: The model weights are publicly available, allowing developers to download, inspect, and fine-tune them.

Q: Can I run Laguna XS.2 on my local machine? A: Yes, Laguna XS.2 is designed to run on-device with only 3B active parameters.

Q: How do I access Laguna M.1? A: Laguna M.1 is available via API, for example through OpenRouter.

Q: What is the pool CLI? A: It is a terminal-based tool that provides an interface to interact with Laguna models, including features like model selection, tool grouping, and keyboard shortcuts.

Newsletter

Join the Community

Subscribe to our newsletter for the latest news and updates