Prerequisites

Install Ollama

Download and install Ollama from ollama.ai/download.

After installation, start the Ollama server:

# Option A: open the Ollama desktop app (macOS / Windows)
# Option B: run from the terminal
ollama serve

Verify it's running:

curl http://localhost:11434
# Expected: "Ollama is running"

(Optional) Log in to Ollama Cloud

Cloud models are hosted by Ollama and require authentication:

ollama login

Follow the prompts in your browser. Once logged in, Cloud models appear in the sidebar's Cloud Models panel.

Install GitHub Copilot Chat

The extension integrates with the GitHub Copilot Chat extension. Make sure it is installed and signed in.

Install Opilot

Search for "Opilot" in the VS Code Extensions panel, or run:

code --install-extension selfagency.opilot

The Ollama activity bar icon (🦙) will appear automatically on next startup.

---

Step 1: Pull Your First Model

Click the Ollama icon in the activity bar to open the sidebar. Navigate to the Library panel, find a model, and click the Pull (⬇) button on any variant.

Recommended starter models:

Use Case	Model	Size
General chat	llama3.2:3b	~2 GB
Code chat (fast)	qwen2.5-coder:7b	~5 GB
Code completions	qwen2.5-coder:1.5b	~1 GB
Vision	llava:7b	~5 GB
Reasoning	deepseek-r1:7b	~5 GB

While a model is pulling you'll see streaming progress inside a VS Code notification. You can continue working — pulls happen in the background.

---

Step 2: Start a Conversation

Via the Model Picker

1. Open GitHub Copilot Chat (Ctrl+Alt+I / Cmd+Alt+I)
2. Click the model selector dropdown in the chat input
3. Select any model prefixed with 🦙 Ollama
4. Start chatting normally

The model picker shows all locally cached Ollama models alongside the standard Copilot models.

Via the @ollama Chat Participant

Type @ollama at the beginning of a Copilot Chat message to direct it straight to your local Ollama instance:

@ollama explain the design of this TypeScript module

The @ollama participant is sticky — once you invoke it in a thread, subsequent messages in the thread continue using it automatically.

→ See @ollama Chat Participant for the full guide.

---

Step 3: (Optional) Enable Inline Completions

To get real-time code suggestions as you type, configure a completion model:

1. Open Settings (Ctrl+, / Cmd+,) and search for ollama.completionModel
2. Enter a small, fast model name (e.g. qwen2.5-coder:1.5b)
3. Make sure ollama.enableInlineCompletions is true

Smaller quantized models (1–3 B parameters) respond quickly enough for a smooth completion experience. See Inline Completions for tuning tips.

---

Step 4: (Optional) Create a Custom Modelfile

Modelfiles let you define a custom model persona, system prompt, and inference parameters. Click the + button in the Modelfiles sidebar panel to launch the wizard.

→ See Modelfile Manager for the full guide.

---

Remote Ollama Instances

To connect to a remote Ollama server, update the ollama.host setting:

1. Open Settings and search for ollama.host
2. Enter your remote URL (e.g., http://my-server:11434)
3. If the server requires authentication, run "Ollama: Manage Ollama Auth Token" from the Command Palette to store your bearer token securely

→ See Settings for all available configuration options.