Skip to content
Copilot for Obsidian

Model Sources and BYOK

Copilot V4 can get models from a Copilot plan, your own provider key or local endpoint, or an agent account you already use. These options are separate. Adding a key does not change the models supplied by Claude Code or Codex.

Choose a model source

Model sourceQuick ChatAgent Chat
Copilot-hosted modelsYesopencode
Your API key or endpoint (BYOK)Yesopencode, when the provider is shown there
Models reported by opencodeNoopencode
Claude Code accountNoClaude
Codex accountNoCodex

Models reported by opencode are routed to their backing provider. Free opencode Zen models show a warning because that provider may log or train on prompts; review its terms before sending sensitive content.

Copilot-hosted models

Lite, Plus, and included Supporter access can provide Copilot-hosted models. You do not need an API key from an AI provider. Enter your license under Settings → Copilot → Basic → Copilot License, then select Apply.

Licensed models can appear in both places:

  • Basic → Agents → Quick Chat for regular Copilot chat.
  • Basic → Agents → opencode for Agent Chat.

They do not appear under Claude or Codex, because those agents use their own accounts and models. The available Copilot lineup changes over time, so use the model lists in settings as the source of truth.

Copilot-hosted models are cloud services, not local models. Brevilabs’s backend and its vetted enterprise model providers process the full request. Copilot’s privacy policy says request content is processed transiently, not retained, and not used for training. See the privacy policy for details.

Bring your own key (BYOK)

BYOK lets you connect Copilot directly to an AI provider, a compatible gateway, or a model server on your computer. For cloud providers, usage and billing stay with that provider.

  1. Open Settings → Copilot → BYOK.
  2. Click Add a provider.
  3. Choose a provider, Ollama, LM Studio, or Add a custom provider.
  4. Enter the API key and Base URL when required.
  5. Select or enter at least one model, optionally click Test, then click Save.

The provider list and model catalog are loaded in the app, so this guide does not keep a fixed provider or model count. If the endpoint cannot list its models, you can enter a model ID yourself.

New chat models are enabled for Quick Chat and opencode by default. You can curate each list independently:

  • Basic → Agents → Quick Chat controls Quick Chat models and its Default model.
  • Basic → Agents → opencode controls the models opencode can use and its Default model.

For enabled, routable models, Copilot passes the saved key and any custom endpoint override to opencode when it starts. A model that opencode cannot route is left out of the opencode list, but may still work in Quick Chat.

Local and Custom Endpoints

Inside BYOK → Add a provider, choose Ollama or LM Studio from the Self Host group to use their local OpenAI-compatible servers. This does not require a Self-Host license. Their API key is optional, and Copilot fills in the usual local URL:

  • Ollama: http://localhost:11434/v1
  • LM Studio: http://localhost:1234/v1

Start the local server before testing the provider. For another compatible service or proxy, choose Add a custom provider and enter its base URL and model ID. The API key is optional; add one when the endpoint requires authentication. Model discovery uses the endpoint’s /v1/models response, but you can always enter an exact model ID before searching the discovered list.

If Test succeeds but Quick Chat cannot send a message, edit the provider and turn on Enable CORS. Responses then appear after completion instead of streaming token by token. For LM Studio, you can enable CORS in LM Studio to keep streaming instead.

Claude Code and Codex accounts

Claude and Codex are Agent Chat backends, not BYOK providers:

  • Claude inherits authentication and models from the Claude Code CLI. Its usage follows your Claude Code account or CLI environment.
  • Codex inherits authentication and models from the Codex CLI through the Codex ACP adapter. Its usage follows your OpenAI account or ChatGPT plan.

Copilot discovers the models reported by each agent and lets you enable them under Basic → Agents → Claude or Basic → Agents → Codex. Those models stay inside their respective agents; they do not become Quick Chat or opencode models.

An OpenAI API key is different from a ChatGPT subscription, just as an Anthropic API key is different from a Claude Code subscription. Use BYOK for API access and the matching agent setup for subscription access.

Where keys are stored

Copilot stores provider API keys and the Copilot license key in this device’s Obsidian Keychain, not in the vault’s data.json. The Keychain is device-specific, so syncing a vault does not copy its credentials to another computer.

To remove stored credentials, open Settings → Copilot → Advanced → API Key Storage and click Delete All Keys.