Welcome to CodeGround Offline AI

Run LNK Official's CodeGround-1.7 model locally on your machine. Write code, build apps, and debug directly from your terminal. 100% private, offline, and free from limits.

Model Registry
CodeGround-1.7
Format / Engine
GGUF / Ollama
Privacy level
Local (No Telemetry)
Architecture
Qwen 2.5 Coder
1

Download CodeGround-1.7 GGUF Model

Fetch LNK Official's offline weights directly via the Hugging Face CLI or secure direct download.

Via CLI (Fastest)

Open PowerShell/CMD and download using the Hugging Face CLI command:

PowerShell
hf download hf://lnkofficial/CodeGroud-1.7/CodeGroud-1.7.gguf

Direct Download

Download the weights directly to your computer using your web browser.

Download GGUF (4.7 GB)
2

Install Ollama for Windows

Ollama handles offline inference, providing a native local server for CodeGround.

Download Installer

Grab the official Windows installer and complete the setup wizard.

Download Ollama for Windows

Verify Installation

Ensure Ollama is running by opening a terminal window and running:

PowerShell
ollama --version
3

Create Modelfile & Run Model

Define a local model structure in Ollama and point it to the downloaded GGUF file.

Create a file named Modelfile (no file extension) in the directory where your CodeGroud-1.7.gguf is saved, and insert:

Modelfile
FROM ./CodeGroud-1.7.gguf
PARAMETER temperature 0.3
PARAMETER num_ctx 8192
SYSTEM "You are CodeGround, a state-of-the-art agentic AI coding assistant created by LNK Official. You help users build, debug, and optimize software directly in their local environment."

Now, build the model and start it using your PowerShell or Command Prompt:

PowerShell
# Build the local model
ollama create codeground -f Modelfile

# Launch CodeGround offline chat CLI!
ollama run codeground
4

Code Directly in CLI (Aider Agent)

Run Aider to allow the local model to read and modify your project files directly.

Install Aider (a terminal-based AI coding partner) using pip and run it using the local Ollama backend:

PowerShell
# Install Aider CLI agent
pip install aider-chat

# Launch Aider connected to local CodeGround model
aider --model ollama/codeground
1

Download CodeGround-1.7 GGUF Model

Fetch LNK Official's offline weights via huggingface-cli or download manually.

Terminal CLI

Download using the Hugging Face CLI command inside macOS Terminal:

Bash
hf download hf://lnkofficial/CodeGroud-1.7/CodeGroud-1.7.gguf

Direct Download

Download the weights directly to your computer using your web browser.

Download GGUF (4.7 GB)
2

Install Ollama for macOS

Ollama provides simple background compilation and runtime for GGUF models on Apple Silicon.

Download App

Download the official macOS zip file, unzip, and drag it into Applications.

Download Ollama for macOS

Verify Installation

Verify that Ollama command-line utilities are registered properly:

Bash
ollama --version
3

Create Modelfile & Run Model

Establish a local entry point in Ollama and point it to the downloaded GGUF file.

Create a file named Modelfile in the directory where your CodeGroud-1.7.gguf is saved, and insert:

Modelfile
FROM ./CodeGroud-1.7.gguf
PARAMETER temperature 0.3
PARAMETER num_ctx 8192
SYSTEM "You are CodeGround, a state-of-the-art agentic AI coding assistant created by LNK Official. You help users build, debug, and optimize software directly in their local environment."

Use macOS Terminal to build and execute the local coding assistant:

Bash
# Build the local model
ollama create codeground -f Modelfile

# Start model in interactive CLI
ollama run codeground
4

Code Directly in Terminal (Aider Agent)

Employ Aider to perform direct code changes within your project directory.

Install the terminal-based AI coding partner via pip and trigger it with the Ollama model:

Bash
# Install Aider CLI agent
pip3 install aider-chat

# Run Aider with local model
aider --model ollama/codeground
1

Download CodeGround-1.7 GGUF Model

Fetch the weights through CLI or direct wget command.

Terminal CLI

Download using the Hugging Face CLI command inside Linux shell:

Bash
hf download hf://lnkofficial/CodeGroud-1.7/CodeGroud-1.7.gguf

Wget Direct Download

Download directly to your current working directory using wget:

Bash
wget https://huggingface.co/lnkofficial/CodeGroud-1.7/resolve/main/CodeGroud-1.7.gguf
2

Install Ollama for Linux

Execute the official one-liner script to download and install Ollama on Linux distributions.

Installation Script

Run the official installation curl script:

Bash
curl -fsSL https://ollama.com/install.sh | sh
3

Create Modelfile & Run Model

Compile a custom configuration to enable native performance.

Create a file named Modelfile in the directory containing the downloaded GGUF file:

Modelfile
FROM ./CodeGroud-1.7.gguf
PARAMETER temperature 0.3
PARAMETER num_ctx 8192
SYSTEM "You are CodeGround, a state-of-the-art agentic AI coding assistant created by LNK Official. You help users build, debug, and optimize software directly in their local environment."

Create and boot the model from Linux Terminal:

Bash
# Build the model in Ollama
ollama create codeground -f Modelfile

# Launch interactive shell
ollama run codeground
4

Launch CLI AI Coding Partner (Aider)

Run Aider to allow CodeGround to modify project files directly.

Install Aider chat agent globally or in your virtual environment and hook it up to Ollama:

Bash
# Install Aider using pip3
pip3 install aider-chat

# Start coding with local CodeGround weights
aider --model ollama/codeground
bash - codeground@offline-agent

Frequently Asked Questions

What are the system requirements for CodeGround-1.7?
Since CodeGround-1.7 is based on the highly optimized Qwen 2.5 Coder architecture (usually 7B parameter size), it runs perfectly on hardware with at least 8GB of RAM. It leverages Apple Silicon (M1/M2/M3) unified memory dynamically, and runs on Nvidia or AMD GPUs via CUDA or ROCm.
How do I use this inside VS Code or Cursor?
You can use the Continue.dev extension for VS Code, JetBrains, or Cursor. In your ~/.continue/config.json, add the model settings pointing to Ollama:

"models": [ { "title": "CodeGround", "provider": "ollama", "model": "codeground" } ]
Why run CodeGround locally?
Running local AI means your source code never leaves your computer, ensuring 100% compliance with strict privacy guidelines. Additionally, there are zero token costs, no daily limits, and it works perfectly even when you have no internet access.