Everything About GIDE

GIDE (GenerativeIDE) is the world's first model-agnostic, offline-capable AI-powered code editor — your AI Coding Command Center. It enables developers to write, debug, refactor, and ship software using the model of their choice: run local models on-device at zero token cost, or connect Claude, GPT, or Gemini with your own API key. In local mode, your code never leaves your machine; in cloud mode, requests go only to the provider you choose.

GIDE stands for GenerativeIDE — a next-generation Integrated Development Environment powered by generative AI. The name reflects its core mission: bringing generative AI capabilities directly into your development workflow, locally and privately.

GenerativeIDE is the company and product behind GIDE. It builds privacy-first, model-agnostic AI developer tooling. GIDE (GenerativeIDE) is the flagship product — a model-agnostic AI code editor built on Visual Studio Code where local models run entirely on your machine, and cloud models via BYOK send only to the provider you choose.

Yes. GIDE is built on a fork of code-oss (the open-source VS Code foundation), giving you the same familiar interface, extension ecosystem via Open VSX, language support, themes, and keybindings you already know — plus native offline AI capabilities built in.

GIDE is built on the open-source VS Code foundation (code-oss). The GIDE-specific AI integrations and infrastructure are proprietary. The app is free to download, and new users get a 30-day free trial on any paid plan.

GIDE bundles a local AI model (Ornith 9B, optimized with llama.cpp) directly into the application. When you interact with the AI in local mode — asking questions, generating code, or refactoring — all inference runs on your local GPU or CPU, so no API calls are made and no data leaves your device. If you opt into a cloud model via BYOK, requests go only to the provider whose API key you supplied.

GIDE is model-agnostic. Locally it runs Ornith 9B — the default bundled model — via llama.cpp, and supports Qwen, Gemma, and DeepSeek as downloadable models. You can also bring your own model in GGUF format, or connect Anthropic Claude, OpenAI GPT, or Google Gemini with your own API key.

GIDE supports Windows 11 (x64), macOS 13.0 Ventura or later (Apple Silicon M1/M2/M3), and Linux (available as .deb for Ubuntu/Debian, .rpm for Fedora/RHEL, and .AppImage for any distro). All three platforms are supported as of version 1.105.4.

On Windows: Windows 11, Intel Core i7 or AMD Ryzen 5000 Series, 16GB RAM, 25GB storage, and a GPU with 12GB+ VRAM (NVIDIA RTX 3060 12GB or better) to run the bundled Ornith 9B — 8GB cards will load smaller models such as DeepSeek R1 7B, but not Ornith 9B at a usable context. On macOS: Apple Silicon M2 or newer with 18GB+ unified memory recommended for the bundled model, 25GB storage. On Linux: similar to Windows GPU requirements.

GIDE is compatible with most VS Code extensions available on the Open VSX marketplace. Extensions that rely on proprietary Microsoft services (like GitHub Copilot or Live Share) may not work, but the vast majority of popular extensions for languages, themes, and tools are fully supported.

GIDE inherits VS Code's language support and can work with virtually any programming language. The AI features are particularly strong with Python, JavaScript, TypeScript, Java, C++, Go, Rust, and other popular languages, which the bundled Ornith 9B model handles particularly well.

Yes, fully — and you can also connect cloud models when you want. In local mode, all AI processing happens on your machine and GIDE never sends your code, prompts, or any AI-related data to external servers; the only operations needing internet are login authentication (OAuth) and update checks, and even those are optional. If you opt into a cloud model via BYOK, only the prompts you send go to that provider using your own API key.

Not in local mode — your code never leaves your machine. GIDE's AI features — chat, autocomplete, refactoring, test generation — all run locally on your hardware using a bundled model, with no telemetry pipelines for code content. Local AI features work fully disconnected, and you can verify this by running GIDE with your network disconnected. Cloud features require internet: if you connect a cloud model via BYOK, the prompts you send go only to the provider you chose.

Yes. In local-only mode, all AI processing happens on-device, making GIDE well-suited for environments with strict data security requirements — including enterprise, government, and defense contexts where code must not leave the device or network perimeter. With no cloud dependencies for local AI features, it is air-gap compatible, and cloud models can be disabled entirely for air-gapped deployments.

GIDE collects only what is necessary for account authentication (email address via OAuth) and for checking software updates. It does not collect, transmit, or store your code, prompts, AI outputs, or development activity. If you choose a cloud model via BYOK, the prompts you send go to that provider under their terms — but GIDE itself never routes your code through its own servers. See the full Privacy Policy at generativeide.com/privacy for details.

Yes. GIDE is designed for enterprise use, particularly in security-sensitive industries. In local mode, AI runs entirely on-device, so it fits naturally into environments where code must remain on-premises — and teams that want frontier-model power can enable cloud models via BYOK where policy allows. Contact us at generativeide.com/enterprise-contact for custom enterprise pricing and support.

Yes. The Enterprise plan is custom-priced, starting at $199.99 per seat per month for a minimum of 10 seats, and adds SSO/SAML, dedicated onboarding, custom compliance reporting, an air-gapped deployment option, and a 1-day support SLA. Contact the team via generativeide.com/enterprise-contact to discuss your organization's needs.

Yes. GIDE can be used for both personal and commercial projects. There are no restrictions on the type of software you build with it.

Cloud-only tools send your code to their servers for AI processing and lock you into their model. GIDE is model-agnostic: run local models entirely offline so your code never leaves your device with zero per-token charges, or connect Claude, GPT, or Gemini with your own API key when you want a frontier model. GIDE is the privacy-first, flexible alternative, on a flat subscription with a 30-day free trial for new users.

Cursor is a cloud-based AI code editor that routes your code and prompts through external AI APIs, so your code leaves your machine and you pay per-token with no offline option. GIDE can run the AI model locally — zero internet required for local AI features, zero token costs, complete code privacy — and it also lets you connect Claude, GPT, or Gemini with your own API key when you want a cloud model, under the same safety controls.

Windsurf (by Codeium) is a cloud-dependent AI IDE. Like Cursor and Copilot, it always sends code context to cloud servers for inference. GIDE is fundamentally different: it can run AI inference on your local GPU/CPU using a bundled model, so no data leaves your machine and there are no ongoing AI costs — and when you want a frontier model, you can connect one via your own API key rather than being locked to a vendor's cloud.

GIDE wins on four dimensions: (1) Privacy — in local mode your code never leaves your machine, and even with cloud BYOK it goes only to the provider you choose, never through GIDE's servers; (2) Cost — no per-token fees and no usage limits on local AI, on a flat subscription; (3) Flexibility — run local models offline or connect Claude, GPT, or Gemini with your own key, with the same agentic harness either way; (4) Reliability — local mode works offline, so API outages don't affect your workflow. For local inference you need a capable GPU for best results, while cloud-only mode has no GPU requirement.

GIDE has three paid plans: Developer at $19.99/month (or $199.99/year), Pro at $99.99/month (or $995/year, with per-seat team billing for 2 to 50 members), and Enterprise with custom pricing from $199.99 per seat per month. New users get a 30-day free trial on any paid plan. Local AI runs on your own hardware, so there are no per-token charges on any plan; if you use a cloud model via BYOK, you pay that provider directly. See generativeide.com/pricing.

Yes. New users get a 30-day free trial on any paid plan — it runs for 30 days or until you purchase a plan, whichever comes first. You are not charged during the trial period; if you purchase a plan, billing starts immediately and the trial ends. During the trial you get full access to GIDE's features, including AI Chat, Autocomplete, and Agentic Coding workflows.

Not for local models. Since GIDE runs local models on your own hardware, there are no token costs or API fees for local AI — you pay nothing per completion, per chat message, or per agentic workflow run, with no usage limits. If you connect a cloud model via BYOK, you pay that provider directly at their rates using your own API key; GIDE never marks up or resells tokens.