🛠️ Are Copilots Local Yet?

Current trends and state of the art for using open & local LLM models as copilots to complete code, generate projects, act as shell assistants, automatically fix bugs, and more.

📝 Help keep this list relevant and up-to-date by making edits!

Table of Contents

• Summary
• Background
• Editor Extensions
• Tools
• Chat Interfaces
• Models
• Datasets
• Misc Tools
• Suggested Setup
• History
• Stats

📋 Summary

Local Copilots are now fully functional, although with output quality still not on par with those offered by cloud-based services like GitHub Copilot.

This document is a curated list of local Copilots, shell assistants, and related projects. It is intended to be a resource for those interested in a survey of the existing tools, and to help developers discover the state of the art for projects like these.

📚 Background

In 2021, GitHub released Copilot which quickly became popular among devs. Since then, with the flurry of AI developments around LLMs, local models that can run on consumer machines have become available, and it has seemed only a matter of time before Copilot will go local.

Many perceived limitations of GitHub's Copilot are related to its closed and cloud-hosted nature.

As an alternative, local Copilots enable:

• 🌐 Offline & private use
• ⚡ Improved responsiveness
• 📚 Better project/context awareness
• 🎯 The ability to run models specialized for a particular language/task
• 🔒 Constraining the LLM output to fit a particular format/syntax.

🧩 Editor Extensions

Editor extensions used to complete code using LLMs:

Name	Editor	⭐	Released	Notes
GitHub Copilot	VSCode, vim	9125	2021-6-29	The GitHub Original, not local or open-source.
Cursor	VSCode	27112	2023-3-14	Fork of VSCode, not open-source
Fauxpilot	VSCode	14645	2022-9-3	Early local PoC. Stale?
Tabby	VSCode, vim, IntelliJ	29074	2023-9-30	Completes the cursor selection
turbopilot	VSCode	3818	2023-4-10	Completions with FIM support, inspired by fauxpilot
HuggingFace-vscode	VSCode	1255	2023-6-19	Fork of Tabnine, supports Starcoder
localpilot	VSCode	3369	2023-10-2	Utility for easily hosting models locally, for use with official Copilot extension using custom API endpoint.
StarcoderEx	VSCode	101	2023-5-5	Completes the cursor selection
WizardCoder-VSC	VSCode	145	2023-6-19	PoC, article available
KoboldAIConnect	VSCode		2023-10-7	Copilot clone using local KoboldAI backend
gen.nvim	vim	1323	2023-10-1	Edit selection using custom prompts
uniteai	VSCode, emacs, lsp	309	2023-8-27
Privy	VSCode	916	2024-1-8	A privacy-first coding assistant.
twinny	VSCode	3279	2024-1-24	The most no-nonsense locally hosted AI code completion plugin for VS Code
continue		21966	2023-5-24	VSCode extension with chat, autocomplete, and actions.

🛠️ Tools

Tools that try to generate projects/features from specification:

Name	⭐	Released	Notes
gpt-engineer	52940	2023-6-6	Specify what you want it to build, the AI asks for clarification, and then builds it.
gpt-pilot	32250	2023-7-18	Very similar to gpt-engineer
aider	25618	2023-6-8	AI pair programming in your terminal, works well with pre-existing, larger codebases
rift	3051	2023-6-20	VSCode extension. Lets you write code by chatting, makes your IDE agentic, AI engineer that works alongside you.
mentat	2583	2023-7-25	Mentat coordinates edits across multiple locations and files.
clippinator	364	2023-4-15	Uses a team of agents to plan, write, debug, and test
Refact.AI	1660	2023-10-06	Full self-hostable code completion, chat and training service, complete with VSCode extension.
LocalCompletion	27	2023-11-15	Inline completion with support for any OpenAI compatible backend

🗨️ Chat Interfaces

Chat interfaces with shell/REPL/notebook access. Similar to/inspired by ChatGPT's "Advanced Data Analysis" feature (previously "Code Interpreter").

Name	⭐	Notes
open-interpreter	57982	open-source, locally running implementation of OpenAI's Code Interpreter
gptme	3131	Supporting open models. Developed by me, @ErikBjare
octogen	256	Local Code Interpreter executing in Docker environment.
terminal-x	34	Very early prototype that converts natural language into shell commands, unmaintained since Sept. 2021
DODA	>50	Electron based GUI for a local OpenAI Dev Assistant

🤖 Models

Models relevant for local Copilot-use. Ordered by most recent first.

Name	Size	Languages	⭐	Released	Notes
Phind CodeLlama v2	34B	Many	829	2023-8-27
WizardCoder-Python	7/13/34B	Python	765	2023-8
CodeLlama	7/13/34B	Many	16165	2023-8
WizardCoder	15B	80+	750	2023-6	Fine-tuning of Starcoder
replit-glaive	3B	1?	88	2023-7	Small model fine-tuned on high-quality data with impressive performance.
Starcoder	15B	80+	7351	2023-5
replit-v1-3b	3B	20+	724	2023-5
SantaCoder	1.1B	Python, Java, JavaScript	331	2023-4	Tiny model selectively trained on 3 languages from 'The Stack'
Qwen 2.5 Coder	32b	92 different languages	3998	2024-11
Deepseek R1	671B	Many	3052	2025-01

Note: due to the pace of new model releases, this section is doomed to be out of date.

📚 Datasets

Datasets relevant for training models.

Name	Size	Languages	⭐	Released	Notes
The Stack	3TB/6TB	358	760	2022-10	Excludes weak-copyleft licenses (MPL, LGPL, EGL) since v1.1

Tools

Misc relevant useful tools.

Name	⭐	Released	Notes
ollama	111009	2023-8-27	Easily get up and running with large language models locally.

Suggested setup

As you can see above there are many options for models and editor extensions. If you use VS Code or JetBrains and want to get started straight away you can use the following setup:

1. Install LM Studio.
2. Install Continue.dev extension.
3. Download one or several models in LM Studio. As of January 2025, Qwen 2.5 Coder is a good choice for autocomplete and Deepseek R1 is a good choice for chat. Depending on your hardware you'll have to experiment with which model size and quantization level gives you sufficient speed. For example on a Macbook Pro M2 with 32GB RAM, Qwen2.5-Coder-7B-Instruct-Q4_K_M works well for autocomplete and DeepSeek-R1-Distill-Qwen-14B-Q4_0 works well for chat.
4. Go to the Developer tab in LM Studio and start the server.
5. Configure Continue.dev extension with by adding your selected models. For example:

   {
       "models": [
           {
           "apiBase": "http://localhost:1234/v1/",
           "title": "Deepseek R1",
           "model": "bartowski/deepseek-r1-distill-qwen-14b",
           "provider": "lmstudio"
           }
       ],
       "tabAutocompleteModel": {
           "provider": "lmstudio",
           "apiBase": "http://localhost:1234/v1/",
           "title": "Qwen 2.5 Coder",
           "model": "qwen2.5-coder-7b-instruct"
       },
   }
   

📰 History

• 🐦 Tweet announcing this repo

📈 Stats

Stargazers over time: