The Ollama sidebar is your command center for model lifecycle management. Click the 🦙 icon in the activity bar to open it.

Local Models Panel

Shows all models that have been pulled to your machine.

Model Tree

Models are grouped by family (tree view) by default. Each family node collapses/expands to show its variants (e.g., llama3.2 → 3b, 1b).

Toggle to a flat list with the Show as Flat List button (⊞) in the panel header, or back to grouped with Show as Tree (⊞).

Capability Badges

Each model item shows capability indicators:

• 🧠 Thinking / chain-of-thought
• 🛠 Tool calling
• 👁 Vision (image input)
• 🧩 Embedding model

Status Indicators

• Running — the model is currently loaded in Ollama's memory. Shows VRAM used and how long it has been loaded.
• Stopped — the model is installed but not loaded.

Inline Buttons

Right side of each running or stopped model item:

Button	When	Action
▶ Start	Stopped	Load model into memory
⏹ Stop	Running	Unload model from memory
🗑 Delete	Stopped	Remove model from disk

Running models cannot be deleted — stop them first.

Panel Toolbar

Button	Action
🔑	Manage auth token for remote instances
⚙ Model settings	Open per-model parameter controls
🔍 Filter	Type to filter by model name
✕ Clear filter	Remove active filter
⊞ / ⊟	Toggle grouped tree / flat list
🔄 Refresh	Reload local model list
⊖ Collapse all	Collapse all family groups

The list auto-refreshes every 30 seconds (configurable via ollama.localModelRefreshInterval).

Model Settings Webview

Use the ⚙ Model settings button (or Ollama: Open Model Settings) to open a webview with per-model controls for:

• temperature, top_p, top_k
• num_ctx, num_predict
• think, think_budget

Updates are applied immediately and persisted per model.

---

Status Bar Heartbeat

Outside the sidebar, Opilot shows a persistent status bar heartbeat:

• Loading: $(loading~spin) Ollama…
• Online: $(pulse) Ollama or $(pulse) Ollama (N)
• Offline: $(warning) Ollama offline (after 2 consecutive failures)

The tooltip includes host, per-model memory, and processor (CPU/GPU) details for currently running models.

---

Cloud Models Panel

Shows models from Ollama Cloud — premium hosted models that run server-side but appear in the sidebar like local models.

Authentication

Click the Login (👤) button in the panel header to authenticate. This triggers ollama login via the Ollama CLI. Once authenticated, cloud models are fetched automatically.

Inline Buttons

Button	When	Action
🔗	Always	Open the model's page on ollama.ai
▶ Run	Stopped	Activate the cloud model
⏹ Stop	Running	Deactivate the cloud model
🗑 Delete	Stopped	Remove from your account

Cloud model names end with :cloud to distinguish them from local variants.

---

Library Panel

Browses the full Ollama model library. Models are displayed as family nodes with expandable variant children.

Pulling a Model

1. Expand a family to see its variants (e.g., 3b, 7b, 13b, 70b)
2. Click Pull (⬇) on the variant you want
3. A streaming progress notification shows download progress

Variants already installed locally show a ✓ checkmark.

Filtering & Grouping

Button	Action
🔍 Filter	Type to filter by model name
✕ Clear	Remove active filter
⊞ / ⊟	Toggle grouped / flat view
🔄 Refresh	Re-fetch from ollama.ai
⊖ Collapse	Collapse all families
🔗	Open model page on ollama.ai

The library is fetched on startup. Use the refresh button or wait for the scheduled refresh (configurable via ollama.libraryRefreshInterval, default 6 hours).

---

Modelfiles Panel

See the dedicated Modelfile Manager page.