README.md

---
title: Mains AI
emoji: 📖
colorFrom: blue
colorTo: indigo
sdk: gradio
sdk_version: 5.16.0
python_version: "3.10"
app_file: app.py
pinned: false
hardware: cpu-basic
---

# Local Reasoning Chat
A general-purpose interactive chat served locally via `llama-cpp-python`
(quantized GGUF, no external API or GPU required). Two response modes trade
speed for depth on the same CPU-only hardware:

- **⚡ Fast** — Qwen3.5-4B, for everyday chat
- **🧠 Deep reasoning** — Qwen3.5-9B, slower but noticeably stronger (e.g.
  UPSC GS-style analysis)

The system prompt is editable in the UI, so this is a reusable base for any
future LLM task, not just one fixed prompt.