Poolside Unveils Open-Source Laguna XS.2 AI Model
Key Takeaways
- 1Poolside launches Laguna XS.2, a 33B parameter LLM for local coding.
- 2New model encourages community innovation and offline functionality.
- 3Shift may boost U.S. AI autonomy, reducing dependency on foreign tech.
- 4• Shift may boost U.S.
- 5AI autonomy, reducing dependency on foreign tech.
Poolside, a San Francisco-based AI startup, has announced the launch of two new large language models: Laguna M.1 and the open-source Laguna XS.2. Laguna XS.2 is designed for local agentic coding tasks, featuring a 33-billion parameter architecture under an Apache 2.0 license, allowing developers the capability to run it without an internet connection. The flagship Laguna M.1 model, although proprietary, is temporarily available for testing through selected partners. Both models were trained from scratch using Poolside's innovative 'Model Factory' environment, ensuring robust performance tailored for enterprise and government applications.
The introduction of Laguna XS.2 represents a significant shift in the AI landscape, emphasizing local autonomy and community-driven development. By enabling offline capabilities, Poolside positions itself uniquely in the market, allowing organizations to deploy AI solutions in secure environments without reliance on external servers. This strategic move not only aims to meet the needs of high-consequence sectors but also enhances the U.S.’s domestic AI capabilities, potentially decreasing dependency on foreign models in critical applications.