blogProduct

What is GIDE? The AI code editor for local, online and self-hosted models

GIDE is an AI code editor that runs local models on your machine, online models on your own key, or models on your own server. Here's what it is, how it works, and who it's for.

3 min readUpdated

Most AI code editors make one decision for you: where your code goes. GIDE leaves that decision with you. It runs local models on your machine, online models on your own key, or models on your own server, and it gives every one of them the same editor and the same controls.

GIDE in one sentence

GIDE is an AI code editor and agent that works with the model you choose, wherever that model runs, and asks before it writes a file or runs a command.

Where your models can run

Local models

A local model runs on your own hardware. GIDE ships with a default local model, Ornith 1.5 9B, and the Model Hub helps you find others: search Hugging Face or pick from a recommended list, and GIDE chooses the largest version that fits your memory, verifies the download, and tests the model before marking it ready.

You can also point GIDE at any .gguf file you already have. These Imported Models get the same temperature and GPU-layer settings as the ones GIDE ships.

Online models on your own key

Bring your own API key for Anthropic, OpenAI or Google Gemini. Keys are stored locally in your system keychain and are only ever sent to the provider you choose. GIDE ships no key of its own and never proxies your prompts.

With Custom Endpoints, you can also connect any endpoint that speaks the OpenAI or Anthropic wire format, such as a self-hosted vLLM or Ollama server, a corporate gateway, or a regional deployment.

Models on your own server

With VPS & Private Servers, you run a model on a VPS, VPN or tailnet you control and connect to it from GIDE like any other model, with your prompts staying on that network. If you don’t have a model server yet, GIDE writes the setup steps for your computer, and one command on the server installs and starts it.

Private Cloud

Private Cloud is GIDE’s own cloud mode. It works alongside your local and online models, with the same controls.

More than one model at a time

GIDE doesn’t make you pick a single model for a whole session. When memory allows, a second local model stays loaded and switching between them is instant. Type @ to send one message to any connected local or online model without changing your default, and the model you address sees the conversation so far.

The same controls for every model

Whichever model answers, it works inside the same harness:

  • Plan Gate. The agent proposes a plan before it acts.
  • Write Gate. Every file write pauses for you to accept or reject.
  • Command Gate. Shell commands show the exact command and wait for your approval.
  • Folder Trust. A folder has to be trusted before the agent reads or writes inside it.
  • Path Gate. When a tool reaches outside your trusted folder, GIDE asks you instead of quietly refusing or quietly allowing it. Reads and writes are separate consents.

The same tools, audit log and memory apply too. Memory is stored locally, so context carries over when you switch models. For why this harness matters more than which model you pick, see One AI code editor for local and cloud models.

In the editor and in the terminal

The GIDE CLI runs the whole agent in your terminal. It talks to the same local server as the editor, so a session you start in one place is the same session in the other, with the same model, the same memory and the same Write Gate.

# Ask the agent from the terminal
gide -p "add a retry to the fetch helper"

# Pick up where you left off
gide --continue

GIDE also includes a Local API Server: gide apikey create mints a key so any OpenAI-shaped client can use the model already running on your machine.

Who it’s for

GIDE is built for developers and teams who want to decide where their code goes:

  • Teams whose code can’t leave their network. Run local models, or a model on your own server, and keep prompts inside your own environment.
  • Developers who want a frontier model for some of the work. Connect an online model on your own key and use it where it helps, next to a local model for the rest.
  • Anyone who wants the agent on a short leash. Plan Gate, Write Gate, Command Gate, Folder Trust and Path Gate keep you in charge of what the agent changes and runs.

Platforms

GIDE runs on macOS, Windows and Linux. On Linux the app ships as .deb, .rpm, .tar.gz and .AppImage. The CLI ships for macOS on Apple Silicon, Linux x64 and Windows x64. For hardware and OS details, see the system requirements.

Try it

Download GIDE from generativeide.com/download. For everything that has changed release by release, see the changelog.