Serve the model with Ollama or LM Studio, connect Continue or Cline, and you've got a local AI coding assistant inside VS Code — no subscription, no code leaving your machine.
You need three pieces, and it takes about 15 minutes: (1) a local server that hosts the model — [Ollama or LM Studio](/ollama-vs-lm-studio-which-should-you-use-2026); (2) the Qwen3-Coder model itself (the right variant for your GPU); and (3) a VS Code extension that connects the editor to that server — Continue (for chat + autocomplete, the closest thing to Copilot) or Cline (an autonomous agent for multi-file tasks). Wire those together and you have a private, free coding assistant inside VS Code — no subscription, and your code never leaves your machine. Here's the exact setup.
Step by step (LM Studio + Cline)
Here's a concrete path using LM Studio and Cline (the same idea works with Ollama + Continue). First, load Qwen3-Coder in LM Studio and start its local server — by default it listens on port 1234. Then install the Cline extension in VS Code (Extensions panel → search 'Cline' → Install). Open Cline's settings and point it at your local server: set the API Provider to 'OpenAI Compatible', the Base URL to `http://localhost:1234/v1`, and the Model ID to your Qwen3-Coder model (e.g. qwen3-coder-30b). That's it — Cline now runs against your local model, and you can hand it multi-file tasks entirely offline.
# LM Studio + Cline (autonomous multi-file agent)
1. LM Studio → load Qwen3-Coder (30B-A3B or 32B) → Start Server (port 1234)
2. VS Code → Extensions → search "Cline" → Install
3. Cline settings:
API Provider : OpenAI Compatible
Base URL : http://localhost:1234/v1
Model ID : qwen3-coder-30b
# Prefer autocomplete + chat (a Copilot replacement)? Use Continue instead:
# Ollama → `ollama pull qwen3-coder` → install "Continue" ext → select the model
If you'd rather have the Copilot-style experience — inline autocomplete plus a chat panel — install Continue instead of Cline, point it at Ollama (or LM Studio), and pick your Qwen3-Coder model. Many people run both: Continue for everyday autocomplete and chat, Cline when they want the model to autonomously edit across several files.
Three pieces — a local server, the model, and Continue or Cline — turn VS Code into a private coding assistant. · Unsplash
Continue vs Cline — which extension?
They do different jobs, so pick by how you work (or use both). Continue is the direct [Copilot](/qwen3-coder-vs-github-copilot-2026) replacement: inline autocomplete, a chat panel, and inline edits — it's the one to install first, and for most developers it covers 90% of daily use. Cline is an autonomous agent: you give it a task ("add tests for this module," "refactor this across files") and it plans and edits across multiple files on its own — it shines with a bigger model like the 32B on a 24GB card, because agentic work is demanding. A common, powerful setup is Continue for fast autocomplete + a small model, and Cline for heavy multi-file tasks + the 32B — which is exactly the kind of local AI agent workflow that used to require a cloud service. Whichever you choose, the payoff is the same: a capable coding assistant that's private, free, and offline — running Qwen3-Coder on your own hardware, inside the editor you already use. Get the right GPU and the right variant, and it just works.
Quick answers
How do I use Qwen3-Coder in VS Code?
You need three things: a local server to host the model (Ollama or LM Studio), the Qwen3-Coder model itself, and a VS Code extension to connect them. A concrete path: load Qwen3-Coder in LM Studio and start its server (default port 1234), install the Cline extension in VS Code, then in Cline's settings set the API Provider to 'OpenAI Compatible', the Base URL to http://localhost:1234/v1, and the Model ID to your model (e.g. qwen3-coder-30b). For a Copilot-style autocomplete-and-chat experience instead, install the Continue extension and point it at Ollama or LM Studio. The whole setup takes about 15 minutes and runs entirely offline.
Continue or Cline for local coding?
They serve different purposes. Continue is the direct Copilot replacement — it gives you inline autocomplete, a chat panel, and inline edits, and it's the one to install first since it covers most daily coding. Cline is an autonomous agent: you give it a task and it plans and edits across multiple files on its own, which works best with a larger model like Qwen3-Coder 32B on a 24GB GPU. Many developers run both — Continue for fast everyday autocomplete and chat, and Cline for heavier multi-file, agentic tasks. Both connect to a local Ollama or LM Studio server, so your code stays private and there's no subscription.
Can I replace GitHub Copilot with a local model?
Yes. With Qwen3-Coder served locally by Ollama or LM Studio and the Continue extension in VS Code, you get inline autocomplete, chat, and inline edits — the core Copilot experience — running entirely on your own machine for free. Your code never leaves your computer, there's no monthly subscription, and it works offline. The main requirements are a capable GPU (24GB is the sweet spot for the 30B-A3B or 32B variants) and a few minutes of setup. For autonomous multi-file tasks beyond what Copilot does, add Cline. It's a genuine replacement for privacy-conscious or cost-conscious developers, provided you have the hardware.
