Tools
Inside the Model Factory — Eiso Kant, Poolside AI
Poolside AI’s co‑CEO Eiso Kant explains that a small, top‑research team built a model‑factory capable of training Laguna S, a 118‑billion‑parameter mixture‑of‑experts model that outperforms the ~1 tri
Poolside AI’s co‑CEO Eiso Kant explains that a small, top‑research team built a model‑factory capable of training Laguna S, a 118‑billion‑parameter mixture‑of‑experts model that outperforms the ~1 trillion‑parameter open‑weight model released by Thinking Machines. The new Laguna S 2.1, highlighted in Poolside’s detailed technical report and a deep dive by Latent Space, shows significant performance gains while underscoring the current debate over open vs. closed‑source AI across the U.S. and China.
Source: Latent Space | 2026-07-23