
Poolside's family of open Mixture-of-Experts foundation models for agentic coding — XS.2 runs locally, M.1 reaches 72.5% on SWE-bench Verified.
Poolside's family of open Mixture-of-Experts foundation models for agentic coding — XS.2 runs locally, M.1 reaches 72.5% on SWE-bench Verified.
Laguna is a family of open Mixture-of-Experts (MoE) foundation models from Poolside built specifically for agentic coding and long-horizon software work. The lineup launched on April 28, 2026 with two models: Laguna XS.2, a 33B-total / 3B-activated model released under the Apache 2.0 license that is compact enough to run locally on a Mac with 36 GB of RAM via Ollama, and Laguna M.1, a 225B-total / 23B-activated model trained from scratch on 30T tokens. On benchmarks, XS.2 scores 68.2% on SWE-bench Verified and M.1 reaches 72.5%, with strong SWE-bench Pro results as well. Poolside also ships 'pool,' a lightweight terminal-based coding agent, and a dual Agent Client Protocol (ACP) client-server — the same environment it uses internally for agent reinforcement-learning training and evaluation.

Laguna by Poolside is a suite of Mixture-of-Experts foundation models designed for agentic coding. The XS.2 model operates locally, while the M.1 model achieves a notable performance score of 72.5% on the SWE-bench Verified benchmark, showcasing its advanced capabilities in coding tasks.
Laguna by Poolside is an innovative platform that integrates the Mixture-of-Experts (MoE) model architecture. This architecture allows the model to activate only a subset of its parameters, optimizing performance and resource usage.
Models Overview:
Use Cases:
Technical Features:
Laguna by Poolside operates using two model sizes, XS.2 and M.1, designed for varying latency and capability needs. It employs a mixture-of-experts architecture for efficient token processing, supports local deployment for private coding, and offers high performance on software engineering tasks.
Laguna by Poolside utilizes two distinct models—Laguna XS.2, with 33 billion parameters (3 billion active), and Laguna M.1, featuring 225 billion parameters (23 billion active). These models serve different user requirements, particularly in terms of speed and computational demands.
This innovative architecture allows the model to route each token through a specific subset of experts, which enhances scalability and efficiency. By not engaging all parameters for every task, Laguna can perform complex operations with reduced latency, making it suitable for both simple and intricate coding tasks.
The XS.2 model is compact enough to be deployed on a Mac with 36 GB of RAM via the Ollama platform, under an Apache 2.0 license. This feature enables users to run coding operations locally, ensuring privacy and control over their projects.
Laguna models have demonstrated impressive results on the SWE-bench benchmark, with XS.2 achieving a score of 68.2% and M.1 reaching 72.5%. These scores reflect the models' capabilities in handling software engineering tasks effectively.
Laguna comes bundled with 'pool,' a lightweight terminal-based coding agent designed to assist users in local, agentic coding. This tool enhances coding efficiency by automating repetitive tasks and simplifying complex code generation.
Laguna by Poolside features two model sizes, a mixture-of-experts architecture for efficiency, local deployment capabilities, strong performance on SWE-bench, and a bundled coding agent. These elements make it suitable for various latency and capability needs, enhancing user experience in AI applications.
Laguna by Poolside offers two distinct model sizes to cater to different user requirements. The Laguna XS.2 is designed for lighter applications with a total of 33 billion parameters and 3 billion active parameters, making it ideal for users needing lower latency. In contrast, the Laguna M.1 boasts a total of 225 billion parameters with 23 billion active, targeting more demanding applications that require robust performance.
One of the standout features of Laguna is its mixture-of-experts architecture. This innovative design allows the model to route each token through a selected subset of experts, enhancing efficiency and scalability. This means that rather than engaging the entire model for every token, only a fraction is used, which reduces computational load and increases speed.
The local deployment option for Laguna XS.2 is a significant advantage. It can be run efficiently on a Mac with just 36 GB of RAM, utilizing the Ollama platform under an Apache 2.0 license. This makes it accessible for developers and businesses that may not have extensive cloud resources.
Performance metrics further validate Laguna's capabilities. The XS.2 model achieves an impressive 68.2% accuracy on the SWE-bench, while the M.1 model reaches 72.5%. These scores indicate strong performance in real-world applications, making it a competitive choice in the AI landscape.
Additionally, Laguna includes a bundled coding agent named 'pool.' This lightweight, terminal-based coding agent streamlines coding tasks, making it easier for developers to integrate AI functionalities into their projects quickly.
Laguna by Poolside is designed for developers and researchers who require robust coding capabilities. It caters to users interested in local agentic coding, high-capability code tasks, self-hosted deployments, research and fine-tuning, and benchmarking software engineering performance.
Laguna by Poolside serves a diverse audience, particularly those engaged in software development and research. Here’s a breakdown of its key functionalities:
Local Agentic Coding: This feature allows users to run XS.2 on their laptops. It’s perfect for individuals who prioritize privacy and want to generate and edit code offline. For example, freelance developers can work on projects without the need for internet access, ensuring data security.
High-Capability Code Tasks: With M.1, Laguna is equipped for handling more complex programming challenges. This capability is essential for long-horizon software engineering projects, such as developing scalable applications or intricate algorithms. Companies managing high-stakes projects can rely on Laguna for reliable performance.
Self-Hosted Deployments: By utilizing open weights, users can avoid dependencies on third-party APIs. This is particularly beneficial for organizations that prioritize data integrity and control over their coding environments. For instance, educational institutions or startups can create customized solutions without worrying about API restrictions or costs.
Research and Fine-Tuning: Researchers can adapt permissively licensed weights to tailor their coding workflows. This flexibility allows for innovation and experimentation in developing new software solutions or improving existing applications.
Benchmarking: Laguna provides tools to evaluate agentic coding performance against established benchmarks like SWE-bench Verified and Pro. This is crucial for software engineers looking to measure efficiency and performance metrics, ensuring they meet industry standards.
Laguna by Poolside is completely free to use, making it an accessible tool for anyone looking to create engaging content. Users can enjoy its features without any subscription fees or hidden costs, which encourages creativity and experimentation with no financial commitments.
Laguna by Poolside offers a user-friendly interface designed for creators, influencers, and businesses looking to enhance their digital presence without incurring costs. By being free, Laguna appeals to a wide audience, including students, small business owners, and hobbyists.
Users can access a variety of features such as content creation tools, templates, and design elements that help in producing visually appealing graphics and videos. This makes it an ideal choice for those who may not have the budget for premium content creation tools.
For example, a small business can use Laguna to create promotional materials and social media posts without needing to invest in expensive software. Additionally, the platform's community support allows users to share tips and tricks, further enhancing the overall experience.
To get started with Laguna by Poolside, visit poolside.ai to sign up for an account. Once registered, you can explore its features, customize your settings, and begin utilizing AI tools to enhance your productivity and creativity.
Laguna by Poolside is a cutting-edge AI platform designed to streamline various tasks and enhance productivity. To begin your journey, head over to poolside.ai. The sign-up process is straightforward; simply provide your email address and create a password. After confirming your email, you'll gain access to the dashboard.
Upon logging in, take a moment to familiarize yourself with the user-friendly interface. Laguna offers tools for content creation, data analysis, and project management. Users can customize their workspace according to their preferences, enabling a tailored experience that suits individual needs.
Avoid common pitfalls such as neglecting to customize your settings or skipping the tutorial, as these can hinder your overall experience.
These resources will help deepen your understanding of Laguna's capabilities and provide additional support as you navigate the platform.
Browse by use case: Code Generation
Compare Laguna by Poolside: vs Soup CLI · vs VibeVoice · vs Arena AI: The Official AI Ranking & LLM Leaderboard · vs PromptLayer