Laguna M.1 is a next-generation Mixture-of-Experts (MoE) large language model developed by Poolside, built specifically for AI coding agents, software engineering, and long-running autonomous workflows.
Unlike general-purpose chat models, Laguna M.1 focuses on complex software development tasks, enabling AI agents to understand large codebases, generate and modify code, analyze projects, and execute multi-step engineering workflows.
The model features approximately 225B total parameters with 23B active parameters, using a sparse MoE architecture to deliver high performance while improving inference efficiency through expert routing.
With a 262K-token context window, Laguna M.1 can process large-scale code repositories, technical documentation, enterprise knowledge bases, and complex development tasks that require long-term context understanding.
Laguna M.1 is designed around the future of AI-powered software engineering, with a focus on planning, tool usage, and solving complex development challenges over extended workflows.
The model is released under the Apache 2.0 open-source license, allowing developers to deploy, customize, and build commercial applications with greater flexibility.
Laguna S 2.1
35 Poolside’s next-generation MoE AI model designed for agentic coding, long-context reasoning, and complex software engineering tasks.
Poolside
Comments (0)