---
title: "Yeco LLM Series — Proprietary Italian Models | YecoAI"
description: "Meet YecoAI's proprietary model family: Yeco-PII (278M privacy model), Ender-1 (3B gaming LLM in beta) and Yeco Flash Mini-IT (118M Italian micro-LLM). Updated as the series grows."
url: "https://yecoai.com/en/products/yeco-llm-series"
locale: "en"
published: "2026-09-29"
updated: "2026-09-29"
category: "Model"
---

# Yeco LLM Series

Our family of proprietary Italian models: Yeco-PII, Ender-1 and Yeco Flash Mini-IT. Sovereign, efficient, compliance-ready. This page is updated as the series grows.

Yeco LLM Series is the model backbone of every YecoAI product: privacy models, gaming-domain LLMs and micro reasoning engines — all trained in-house, on our own data, for a fraction of frontier training costs.

### Sovereign by design

Trained in Italy, on our own data pipelines. No third-party API brains — full digital sovereignty for your business.

### Efficiency over brute force

Compact models with rigorous evaluation: Yeco-PII (278M) and Flash Mini-IT (118M) outperform far larger rivals on their verticals.

### Compliance-ready

GDPR and EU AI Act by construction: training data policy-controlled, deployment CPU-only, no data leaving your infrastructure.

## I modelli

Yeco LLM Series è la nostra famiglia di modelli italiani proprietari. Ogni modello ha una pagina research dedicata e un annuncio di lancio — e questa pagina viene aggiornata man mano che la serie cresce.

## Yeco-PII v1.0

Modello da 278M parametri per detection e anonimizzazione di PII. Batte l'OpenAI Privacy Filter (1.5B) su tutte le 6 lingue europee (F1 0,966 vs 0,769). Rilasciato ufficialmente il 10 set 2026. Trial API enterprise e licenza on-prem una tantum disponibili.

## Ender-1 (Beta)

Modello proprietario da 3B addestrato sui dati di EnderDevelopment (utenti free, nel rispetto delle policy). Progettato per generazione di codice e configurazioni nell'ecosistema gaming. In rollout su EnderDevelopment da oggi.

## Yeco Flash Mini-IT v1

Micro-LLM italiano da 118M parametri, addestrato con meno di €15, ~200 tok/s su una CPU consumer. Batte modelli 5× più grandi nei benchmark di logica italiana (BoolQ: 62,0 vs Minerva-350M 60,7). Apache 2.0, soggetto a rielaborazione man mano che la serie evolve.

```json
{
  "@context": "https://schema.org",
  "@id": "https://yecoai.com/en/products/yeco-llm-series#page",
  "url": "https://yecoai.com/en/products/yeco-llm-series",
  "name": "Yeco LLM Series — Proprietary Italian Models | YecoAI",
  "description": "Meet YecoAI's proprietary model family: Yeco-PII (278M privacy model), Ender-1 (3B gaming LLM in beta) and Yeco Flash Mini-IT (118M Italian micro-LLM). Updated as the series grows.",
  "inLanguage": "en",
  "isPartOf": {
    "@id": "https://yecoai.com/#website"
  },
  "@type": "SoftwareApplication",
  "applicationCategory": "Model",
  "operatingSystem": "Web",
  "publisher": {
    "@id": "https://yecoai.com/#organization"
  },
  "author": {
    "@id": "https://yecoai.com/#organization"
  }
}
```

```json
{
  "@context": "https://schema.org",
  "@type": "BreadcrumbList",
  "itemListElement": [
    {
      "@type": "ListItem",
      "position": 1,
      "name": "YecoAI home",
      "item": "https://yecoai.com/en"
    },
    {
      "@type": "ListItem",
      "position": 2,
      "name": "Products",
      "item": "https://yecoai.com/en/products"
    },
    {
      "@type": "ListItem",
      "position": 3,
      "name": "Yeco LLM Series",
      "item": "https://yecoai.com/en/products/yeco-llm-series"
    }
  ]
}
```