הפוסט הזה כתוב באנגלית.

Free, offline Tab with a local model

Ingot's Tab completion can run on a coder model in Ollama on your own computer. It costs nothing, works offline and no code leaves the machine. Here is how to set it up and how Ingot decides when to use it.

Claire Vasseur2 דקות קריאה

A grey Tab suggestion from a local Ollama model in a TypeScript file, with Tab · local in the status bar

Tab completion is the feature you use hundreds of times a day without thinking about it. In Ingot it can run on a small coder model in Ollama on your own computer. Local Tab costs $0, works on a plane, and no code leaves your machine.

Everything else stays the same: the ghost text, the keys, next edits and the status bar.

Set it up once

  1. Install Ollama from ollama.com/download and start it.
  2. Install a coder model, about 1 GB:
ollama pull qwen2.5-coder:1.5b

You can also do step 2 from Ingot: click Tab in the status bar and pick Install a local model…. A terminal opens with the command already typed; press Enter. Tab picks the model up by itself when the download finishes.

The Tab status item reading Tab · local

When Ingot uses the local model

Free local Tab is on by default, with the setting ingot.tab.provider at auto. With auto, Tab uses the local model when Ollama has a coder model installed and one of these is true:

  • privacy mode is on
  • you are signed out
  • your plan has no hosted Tab

Otherwise Tab uses Ingot's hosted model. Want local every time? Set ingot.tab.provider to local. Then, when Ollama is not running, the status bar says so instead of quietly falling back to hosted Tab.

Which model it picks

Leave ingot.tab.localModel empty and Tab picks the best coder model you have installed, preferring speed: Qwen2.5 Coder 1.5B first, then 3B.

Fill-in-the-middle models also see the code after the caret, so their suggestions fit better in the middle of a file. Qwen2.5 Coder, StarCoder2, DeepSeek Coder, Code Llama, CodeGemma and Codestral all do this.

Fast by design

  • Tab waits the same short typing delay as hosted Tab.
  • If you keep typing, the request to Ollama is cancelled at once.
  • Replies are capped at 64 tokens, and recent suggestions are cached in memory.
  • Local Tab does not count against Tab's daily limits.

Private by design

Ollama is always contacted by the Ingot backend, never by a web page. In the desktop app that backend is your computer, so your code never leaves it. That pairs well with privacy mode, which is on by default.

One thing stays hosted: Tab in the agent chat box. While Tab runs locally, the chat box shows no ghost text.

When it does not suggest anything

Hover the Tab item in the status bar, or run Ingot: Why isn't Tab suggesting? from Find Action. It tells you the reason and offers the fix: Open Ollama download page, Install a local model… or Check for Ollama again. The same check is a row in Ingot Health.

The full reference is in Free local Tab.