title: Qwen LLM
created: 2026-07-04
updated: 2026-07-04
type: entity
tags: [apps, llm, qwen, models, inference]
Qwen LLM
The primary language model serving all apps. Runs on Windows behind llama.cpp server, accessible to all nodes via Tailscale.
Details
| Property | Value |
|---|---|
| Host | Windows (100.119.202.68) |
| Port | :8000 |
| Model | Qwen3.6-27B |
| Context | 131K tokens |
| Speed | ~49 t/s on RTX 3090 |
| API | OpenAI-compatible /v1/chat/completions |
| Provider | llama.cpp server |
Server Variants
| Model | Description |
|---|---|
| 27b | High quality reasoning (default) |
| 9b | Fast, low power |
| nemotron | Efficient alternative |
Main Functions
- Text generation and reasoning
- Chat completions via OpenAI-compatible API
- Document analysis and summarization
- Transcription and language processing
Dependencies
Cross-Functionality
- Powers Decksmith presentation content generation
- Powers RentReport market analysis
- Powers Meeting App transcription and AI assistant
- Powers File Processor document analysis
- Powers SimpleChat conversations
- Powers Hermes Agent System — all agent reasoning
- Routed through AI Gateway for model selection