A
Continue logoContinueOffline autocomplete and chat in your editor

Local Coding Copilot in VS Code (Zero Cloud)

setuproll@setuproll
86.0Overall score

Wires a local model into VS Code for both inline autocomplete and chat, keeping your proprietary code entirely on your machine. For developers under strict data rules or anyone who refuses to ship source to a cloud provider.

86.0Score
1.8kVotes
5Components

Install this build

Export
terminal
ollama pull qwen3-coder:30b && ollama pull qwen2.5-coder:1.5b

Components

Model

  • Qwen3 Coder 30B (chat)
  • Qwen2.5 Coder 1.5B (autocomplete)

Stack

  • Ollama
  • Continue extension

Hardware

  • 24GB VRAM GPU or 32GB unified Mac
  • 16GB works for smaller quants

Quantization

  • Q4_K_M for the chat model
  • FP16 for the tiny autocomplete model

How it works

  • Pull a big coder model for chat and a tiny one for tab completion
  • Add both to Continue config as chat and autocomplete roles
  • Edit, ask, and refactor with full repo context offline
  • Swap models per task without leaving the editor

Summary

Wires a local model into VS Code for both inline autocomplete and chat, keeping your proprietary code entirely on your machine. For developers under strict data rules or anyone who refuses to ship source to a cloud provider.

86.0 score 1.8k votes

0 Reviews

Your rating
Sign in to post

Loading discussion...

← All builds