Rafeeq
Download

Docs

Getting started

From installer to first answer, in three steps.

1 Install Rafeeq

Download the installer for your platform from the Download page. If you're not sure which variant to pick, the one with the starter model included is the easiest way to get going.

2 Add a model (standard installer only)

If you installed the standard build, Rafeeq needs at least one .gguf model file before it can chat:

  1. 1Go to huggingface.co and search for a model with GGUF in the name.
  2. 2Download a quantized file — one named like *.Q4_K_M.gguf is a good balance of size and quality.
  3. 3Open Rafeeq, go to the Models screen, and click Open models folder. Drop the file in, then click Refresh.

Models folder, standard installer on Windows

%LOCALAPPDATA%\com.ayoubedahlouli.rafeeq\models

The Models screen and Settings always show the exact path for your install. With the starter-model installer, the folder name ends in .bundled.

3 Start chatting

Load a model from the Models screen, then pick a mode — General, Code help, Document, or RAG — depending on what you're working on.

  • General

    Everyday questions.

  • Code help

    Explain and suggest, never run.

  • Document

    Ask about one file.

  • RAG

    Ask about a folder.

Setting up local RAG

RAG needs a second, small embedding model (not your chat model), for example nomic-embed-text-v1.5. Sideload it the same way as a chat model, open the Knowledge base page, select it as the embedding model and add a folder of documents. Indexing runs in the background and can be paused. Then ask questions in RAG mode.

Verifying it's really offline

Rafeeq makes zero network calls at runtime, by construction. You can test this yourself: block network access at the OS level (firewall rule, disabled adapter, or an air-gapped VM) and Rafeeq will run identically. The app also shows an offline status indicator in the header.

Offline · no network calls

More

For build-from-source instructions and full technical details, see the README on GitHub.