SetuprollSetuproll
A
Aider logoAiderA private Discord assistant with no API bill

Self-Hosted Discord Bot on a Local LLM

setuproll@setuproll
84.0Overall score

A Discord bot wired to a local Ollama model so your server gets an always-on assistant that costs nothing per message and keeps every chat on your own box. For communities and homelabs that want a custom bot without renting tokens.

84.0Score
1.3kVotes
5Components

Install this build

Export
terminal
ollama run qwen3:14b && python bot.py

Components

Model

  • Qwen3 14B (Q4_K_M)
  • Llama 3.3 8B for low VRAM

Stack

  • Ollama
  • discord.py
  • Redis for short-term memory

Hardware

  • 12GB VRAM GPU or 16GB unified Mac
  • Runs fine on a spare desktop

How it works

  • Run Ollama and pull a mid-size chat model
  • discord.py listens for mentions and slash commands
  • Redis keeps the last N turns per channel for context
  • Stream tokens back into the message as they generate

Rules

  • Rate-limit per user to protect the GPU
  • Strip server invites and links from prompts

Summary

A Discord bot wired to a local Ollama model so your server gets an always-on assistant that costs nothing per message and keeps every chat on your own box. For communities and homelabs that want a custom bot without renting tokens.

84.0 score 1.3k votes

0 Reviews

Your rating
Sign in to post

Loading discussion...

← All builds