Apaco
A LoRA adaptation of Qwen 3.5-9B designed to be used locally in disasterous or apocolyptic situations, completely of the grid.
Requirements
minimum 8Gb memory, recommended 16-32Gb
No Wifi or Cellular connection required
operable on both Linux and Windows, support for Local LLM tools planned.
APACO
An offline AI survival system powered by Qwen.
APACO is an experimental survival-focused AI system designed to provide practical knowledge, reasoning, and decision support during emergencies, infrastructure failures, and prolonged disruptions to normal life.
Its primary objective is simple:
Build an AI that remains useful when the internet, cloud services, and modern infrastructure are no longer available.
Rather than attempting to compete with general-purpose AI assistants, APACO focuses on practical survival, self-sufficiency, resource management, and maintaining essential systems with limited resources.
1. Core Objectives
APACO is built around six priorities:
- Offline Intelligence: Run locally without requiring an internet connection or external API.
- Survival Knowledge: Provide practical information covering essential survival disciplines.
- Self-Sufficiency: Help users produce, repair, maintain, and manage essential resources.
- Adaptability: Account for limited supplies, tools, electricity, and equipment.
- Efficiency: Operate on accessible consumer hardware rather than requiring data-center infrastructure.
- Reliability: Prioritize accurate information, source verification, and transparent uncertainty.
APACO is intended to function as an independent knowledge and decision-support system.
2. Model Architecture
APACO uses an open-weight Qwen model as its underlying language and reasoning engine.
| Component | Purpose |
|---|---|
| Qwen 9B | Primary language and reasoning model |
| Local inference | Offline model execution |
| Knowledge base | Stored |
APACO will adapt to Talos based when applicable, but for the time being will use qwen models.