AIRGAP StudioDownload
Install Airgap Studio and set up your offline AI development environment.
License key required
A license key is required to use AIRGAP Studio. If you don't have one yet, get a free 14-day trial.
Standard All-in-One (Recommended)
AIRGAPStudio-Standard-1.0.0.zip
All-in-one package including the IDE, AI engine, 7 user-facing extensions, and the default AI model (Qwen3 8B). Extract the ZIP and run the installer to get started. Recommended for most users.
Includes
- AIRGAP Studio IDE + AI Engine (llama-server + Vulkan GPU acceleration)
- AIRGAP Assistant, Designer, Bridge, UI Library, Autocomplete, License
- Security Shield (SAST scanner, DLP engine, audit logs)
- Qwen3 8B AI model (GGUF, 4.7 GB)
- Default settings and environment configuration
AIRGAP Studio IDE
AIRGAPStudio-Setup-1.0.0.exe
IDE-only installer with the AI engine included. AI models are not bundled — download a model pack separately. Useful for air-gapped environments where files are transferred via USB.
Includes
- AIRGAP Studio IDE
- AIRGAP Assistant, Designer, Bridge, UI Library, Autocomplete, License
- Security Shield (SAST scanner, DLP engine, audit logs)
- AI Engine (llama-server + Vulkan GPU acceleration)
- Default settings and environment configuration
Standard Model Pack (Qwen3:8b)
qwen3-8b-Q4_K_M.zip
Qwen3 8B GGUF model file. Download this if you installed the IDE only. Copy the .gguf file into the models folder of your installation directory.
Includes
- Qwen3 8B model file (GGUF, single file)
- Installation guide (README.txt)
Premium Model Pack (GPT-OSS 20B)
gpt-oss-20b-mxfp4.zip
Coming soon — download not yet available. An advanced MoE (Mixture of Experts) model for agentic coding that activates only 3.6B of 21B parameters for efficient inference. Ideal for complex reasoning and refactoring. Requires a GPU with 16 GB+ VRAM.
Includes
- GPT-OSS 20B model file — coming soon (download not yet available)
- Installation guide (README.txt)
AI Model Comparison
Choose a model based on your use case. Both models can be installed and switched between.
| Spec | Qwen3 8B (Default) | GPT-OSS 20B (Coming Soon) |
|---|---|---|
| Parameters | 8B (Dense) | 21B / Active 3.6B (MoE) |
| Model Size | ~4.7 GB (GGUF Q4_K_M) | ~11.3 GB (GGUF MXFP4) |
| GPU VRAM | Recommended 6 GB+ (min 4 GB, auto CPU fallback) | 16 GB+ |
| Context | Up to 40K tokens (GPU) / 16K (CPU) | 128K tokens |
| Korean | Strong (119 languages) | Supported (English-focused) |
| Best For | Autocomplete, basic chat, lightweight tasks | Agentic coding, complex reasoning, refactoring (coming soon) |
* All models are Apache 2.0 licensed and run 100% locally with no external communication.
System Requirements
Works without a GPU. With a GPU, AI response speed improves 2-5x.
| Spec | Minimum | Recommended |
|---|---|---|
| OS | Windows 10 64-bit | Windows 11 64-bit |
| CPU | Intel i5 / AMD Ryzen 5 | Intel i7 / AMD Ryzen 7 |
| RAM | 16 GB | 32 GB |
| GPU (optional) | Vulkan-compatible GPU | NVIDIA RTX / AMD Radeon (6 GB+ VRAM) |
| Disk Space | 10 GB free | 20 GB free (SSD) |
| Network | Not required (fully offline) | — |
Installation Steps
Simple 3-step installation
Download
Download the Standard bundle (~4.9 GB). It includes the IDE, AI engine, and default model (Qwen3 8B).
Install
Extract the ZIP and run the installer. No admin privileges required — the AI engine runs in the background via Launcher.
Activate License & Start
Launch AIRGAP Studio and enter your license key. AI features activate immediately with full access to all capabilities. Don't have a key yet? Get a free 14-day trial.