---
id: 20260724-T0-02
title: "Poolside AI揭秘模型工厂：小团队如何训练出118B MoE模型"
title_en: "Inside Poolside AI's Model Factory: Small Team Trains 118B MoE"
url: https://ai.daily.yangsir.net/daily/20260724-T0-02
issue_date: 2026-07-24
publish_date: 2026-07-23T05:09:14.000Z
category: insight
source_name: "Latent Space"
source_url: https://www.latent.space/p/poolside
---

# Poolside AI揭秘模型工厂：小团队如何训练出118B MoE模型

Poolside AI联合CEO Eiso Kant接受Latent Space采访，透露其小团队由顶尖研究人员组成，已建成模型工厂并成功训练出Laguna S——一个118B参数的MoE模型，性能超越Thinky约1T参数的开源模型。他表示这仅仅是开始。

## English Version

**Inside Poolside AI's Model Factory: Small Team Trains 118B MoE**

Poolside AI co-CEO Eiso Kant reveals on Latent Space that a small elite team built a model factory to train Laguna S, a 118B MoE model outperforming Thinky's ~1T open-weight model. He calls it just the beginning.

---

**来源**：[Latent Space](https://www.latent.space/p/poolside)

**详情页**：https://ai.daily.yangsir.net/daily/20260724-T0-02

---

*智语观潮 · Daily — https://ai.daily.yangsir.net/llms.txt*