README.md

---
language:
  - cs
  - sk
  - en
  - de

license: apache-2.0

base_model: EuroLLM-9B

quantization: Q8_0

pipeline_tag: text-generation

library_name: llama.cpp

tags:
  - gguf
  - llama.cpp
  - offline
  - offline-first
  - local-ai
  - multilingual
  - ai-runtime
  - webui
  - cli
  - voice
  - speech-to-text
  - text-to-speech
  - privacy
---

# Offline AI Runtime 3.0 – EuroLLM-9B-Q8_0 (GGUF)

Offline AI Runtime 3.0 is a fully local AI runtime environment focused on digital sovereignty, privacy, and user-controlled AI.

Everything runs entirely on your own computer.

**No cloud.**  
**No telemetry.**  
**No tracking.**  
**No subscriptions.**

Powered by **llama.cpp**.

---

# 🖥️ Runtime Interfaces

Offline AI provides multiple ways to interact with local AI:

- Terminal (CLI)
- Modern Web UI
- Voice interaction (Speech-to-Text + Text-to-Speech)

All interfaces use the same local runtime and the same locally running language models.

![Terminal or Web UI](Terminal_WebUI.png)

---

# ✨ What's New in 3.0

Offline AI has evolved from a simple model launcher into a complete local AI runtime platform.

New features include:

- Modern Web UI
- Structured CLI runtime
- Voice conversations (STT + TTS)
- Runtime model switching
- Model lifecycle management
- Model alias system
- Workspace profiles
- Conversation snapshots
- Notes system
- Runtime diagnostics
- Administrative control mode
- Improved modular architecture

Designed for running multiple local AI models inside a unified offline environment.

---

# 🧠 Runtime Architecture

```text
CLI / Web UI / Voice
          │
          ▼
 Offline AI Runtime
          │
 ├── Profiles
 ├── Notes
 ├── Snapshots
 ├── Diagnostics
 ├── Voice
 ├── Model Manager
 └── Runtime Commands
          │
          ▼
      llama.cpp
          │
          ▼
     GGUF Models
```

The Python runtime acts as the orchestration layer while **llama.cpp** provides efficient local inference with real-time token streaming.

---

# 🔧 Technical Information

| Component | Value |
|-----------|-------|
| Base model | EuroLLM-9B |
| Quantization | Q8_0 |
| Format | GGUF |
| Runtime | Offline AI Runtime 3.0 |
| Inference Engine | llama.cpp |
| Recommended RAM | 16 GB |
| Platforms | macOS • Windows • Linux |

This repository distributes an **unmodified quantized GGUF version** of EuroLLM-9B.

The original model has **not been fine-tuned or modified** as part of this project.

---

# 🔐 Project Philosophy

Offline AI explores a different direction for modern AI.

The project demonstrates that:

- AI can work without cloud services.
- AI belongs to its owner.
- Privacy should be the default.
- Users should control their own models.
- Local-first AI is a viable architecture.

Offline AI promotes:

- Digital sovereignty
- Local-first computing
- Transparent system design
- Open AI ecosystems
- Privacy by default

---

# 📄 Model Origin & License

**Base Model**

EuroLLM-9B

**Authors**

EuroLLM Consortium

**Base Model License**

Apache License 2.0

**Runtime Engine**

llama.cpp (MIT License)

**Offline AI Runtime**

© David Káninský

All components are distributed in accordance with their respective licenses.

---

# ⚠️ Disclaimer

Offline AI is an educational and experimental runtime environment.

Generated responses may contain inaccuracies and should not be used as the sole basis for legal, medical, financial, or other high-risk decisions.

Users are responsible for how the software and generated content are used.

---

# 🌍 Project

**Website**

https://OfflineAI.online

Additional domains:

- OfflineAI.cz
- OfflineAI.sk
- OfflineAI.de

**Author**

David Káninský