US startup Poolside debuts its first open-weight model, Laguna XS.2, a 33B-A3B-parameter MoE model, and Laguna M.1, a proprietary 225B-A23B-parameter MoE model
The AI race lately has felt a bit like a game of tennis: first, Anthropic releases a new, pricey state-of-the-art proprietary model …
Lead Source
How this story grew
More
Discussion
TechSnif Coverage
Poolside Drops First Open-Weight AI Model Alongside Beefy Proprietary One
US startup Poolside launches two new MoE models: the open-weight Laguna XS.2 and proprietary Laguna M.1.
Poolside just entered the AI arena with a double release. The US startup unveiled Laguna XS.2, its first open-weight model, packing 33 billion total parameters with 3 billion active in a Mixture of Experts (MoE) architecture. Translation: it's lean and accessible.
Alongside it comes Laguna M.1, a much heftier proprietary model sporting 225 billion parameters with 23 billion active. That's the big gun — kept behind closed doors.
The timing is deliberate. The AI model race has turned into a rapid-fire volley between major players, with new releases dropping constantly. Poolside is positioning itself as a fresh contender with a dual strategy: give the community something to tinker with via open weights, while keeping the flagship locked down.
Two models. Two strategies. One startup betting it can compete with the heavyweights.