---
title: "Home"
description: "Synora builds AI models with open weights by default and researches intelligence that does not depend on training at scale."
url: "https://synora.bg/"
---

# Foundational intelligence, created and trained in Europe.

Synora builds and specialises AI models for cybersecurity, defence, physical systems and enterprise, and releases them with open weights. Weightless Intelligence, our second line of research, pursues intelligence that does not depend on training at scale.

## Technology

Two lines of research: models, inside today's paradigm, and Weightless Intelligence, outside it.

- [AI and models](https://synora.bg/technology#models) — Base, foundation and specialised models, trained in-house and released with open weights by default.
- [Physical AI and autonomous systems](https://synora.bg/technology#physical) — Models for machines that perceive and act, trained in simulation and tested against the physical world.
- [Cybersecurity and security intelligence](https://synora.bg/technology#security) — Specialised models for security analysis: code, logs, network traffic and the systems behind them.

*Featured work*

November 2026 · Announcement

## To the foundations

Why Synora is building its own models, and what it commits to.

[Read the announcement](https://synora.bg/work/to-the-foundations)

## Latest work

[All work](https://synora.bg/work#all)

- [To the foundations](https://synora.bg/work/to-the-foundations) — Announcement, November 2026, Synora

## Initiatives

- [Weightless Intelligence](https://synora.bg/work/weightless-intelligence) — Research into intelligence that does not depend on training at scale, aimed first at autonomy and Physical AI.

*Company*

## An independent company.

Synora is founder-funded, with no outside equity. It sets its own research agenda and carries its own risk.

[About Synora](https://synora.bg/company)
