Copilot for Obsidian
THE Copilot in Obsidian
Last verified:
What is Copilot for Obsidian?
Copilot for Obsidian is an open-source, privacy-focused AI assistant plugin that integrates Large Language Model (LLM) capabilities directly into the Obsidian note-taking application. It allows users to chat with their entire Obsidian vault using numerous LLM providers, enabling semantic search, question-answering, and AI-powered commands based on personal note content. The plugin serves as a
Copilot for Obsidian pricing
Pricing model: Freemium
Free Tier (BYOK - Bring Your Own Key): All core features are free including chat mode, Vault QA (basic), lexical search, custom prompts, commands, Relevant Notes, Composer, and multimedia understanding (images, YouTube, web). Users must provide their own API key from providers like OpenAI, Anthropic, Google, Ollama, or OpenRouter (recommended). Copilot Plus (Paid): Premium tier with advanced agentic AI features. Includes built-in chat model (copilot-plus-flash/Gemini Flash) and embedding models (copilot-plus-small default, copilot-plus-multilingual, copilot-plus-large for Believer Plan). Features: Autonomous Agent Mode with automatic tool calling, time-based queries, multimodal chat context (images, PDFs, YouTube, URLs, web search), Projects Mode, @vault tool (more powerful than Vault QA), and long-term memory for the agent. Two plans: PLUS and BELIEVER (Believer includes copilot-plus-large for complex legal/math content). One-time payment option for self-host forever. 14-day full refund policy, no questions asked. License key obtained from dashboard at obsidiancopilot.com/en/dashboard and entered in Copilot Settings → Basic Settings → Plus License field.
Copilot for Obsidian pros
- Open-source core with privacy-focused design
- 100% local data ownership and storage
- Bring Your Own Model (BYOK) - no vendor lock-in
- Works with OpenAI, Anthropic, Google, Ollama, LM Studio, and OpenAI-compatible models
- Chat with entire vault using Vault QA mode
- Lexical search works out-of-the-box without setup
- Semantic indexing optional for deeper recall across thousands of notes
- Projects mode supports 50+ file types including PDFs, images, Office docs
- Composer allows editing notes directly within chat with one-click changes
- Autonomous Agent Mode with automatic tool calling (Copilot Plus)
- Time-based queries for precision vault search
- Multimodal support: images, YouTube transcripts, URLs, PDFs as context
- Relevant Notes suggestions appear automatically based on semantic similarity
- Custom prompts and commands via command palette or right-click menu
- Plain Markdown files for memory, chat history, and system prompts
- Self-host option for truly private on-device LLM workflows
- Quick Ask floats at cursor for fast interactions
- 100,000+ users trust it - No. 1 downloaded Obsidian AI integration
- Supports fetching webpages, YouTube transcripts, and tweets on demand
- Inline citations from notes and live sources ground every answer
Copilot for Obsidian cons
- Copilot Plus premium features require paid subscription
- BYOK users must obtain and manage their own API keys
- API rate limits and quotas depend on chosen provider
- Image understanding only available in Copilot Plus
- Autonomous Agent Mode only available in Copilot Plus
- Mobile/iPad support is experimental with limited feature parity
- Large PDFs (>10MB) need conversion to Markdown for better parsing
- Token limit errors can occur if max token setting is too high
- Switching embedding models after indexing can break search results
- Copilot Plus backend code is closed-source and proprietary
Frequently asked questions about Copilot for Obsidian
What is Copilot for Obsidian?
Copilot for Obsidian is an Obsidian plugin that allows you to chat with your vault using numerous LLMs. It is an open-source and privacy-focused AI assistant for your personal wiki. It enhances Obsidian with AI capabilities through a dedicated chat interface, semantic search and question-answering based on vault content, AI-powered commands and custom prompts, document understanding and reference capabilities, and advanced AI agent functionality via Copilot Plus.
Is Copilot for Obsidian free?
All features except Copilot Plus are free to use. The free tier requires you to bring your own API key (BYOK) from providers like OpenAI, Anthropic, Google, Ollama, or OpenRouter. Copilot Plus is a paid tier with advanced features including autonomous agent mode, built-in chat model (no API setup required), time-based queries, multimodal support, and Projects Mode.
How do I install Copilot for Obsidian?
Open Obsidian → Settings → Community plugins → Turn off Safe mode (if enabled) → Click Browse → Search for 'Copilot for Obsidian' → Click Install → Click Enable. Once installed, you'll see a chat icon in your left-side ribbon; clicking it opens the chat panel on the right. Before chatting, you need to add your API key in Settings → Copilot Settings → Basic Settings → API Keys.
What AI models can I use with Copilot?
Copilot is model-agnostic with zero lock-in. You can use OpenAI (gpt-4o, gpt-4o-mini), Anthropic, Google, Ollama, LM Studio, or any OpenAI-compatible model. You decide which model to use and can switch anytime. Copilot Plus includes a built-in chat model (copilot-plus-flash/Gemini Flash) and embedding models (copilot-plus-small, copilot-plus-multilingual, copilot-plus-large for Believer Plan) without requiring provider setup.
How does Vault QA mode work?
Vault QA mode lets you chat with your entire vault using a combination of chat and embedding models. It uses lexical search out-of-the-box (no setup required) and optionally adds semantic indexing for deeper recall. Ask questions like 'What are the recurring themes in my research regarding AI and SaaS?' Copilot retrieves relevant notes and provides answers grounded in your content. For Copilot Plus users, the @vault tool in Agent Mode is more powerful than basic Vault QA.
Why isn't Vault search finding my notes?
If Vault QA mode or the @vault tool in Plus isn't finding notes: 1) Ensure you have a working embedding model from your AI provider (e.g., OpenAI), 2) Ensure Copilot indexing is up-to-date, 3) Run 'Force Re-Index' or 'List Indexed Files' from Command Palette to inspect indexed files, 4) Don't switch embedding models after indexing as it can break results. Also check QA settings for inclusion/exclusion filters and raise 'Max Sources' if needed.
What is Copilot Plus Agent Mode?
Agent Mode is a Copilot Plus feature where the AI autonomously calls the right tools without manual commands. Just ask, and Copilot automatically searches the web, queries your vault, analyzes YouTube videos, or edits notes using @vault, @websearch, @youtube, and @composer tools. The agent can also use long-term memory (since v3.1.0) to remember past conversations. Example: 'Research web and my vault and draft a note on AI SaaS onboarding best practices.'
Can Copilot understand images in my notes?
Image understanding is only supported in Copilot Plus. To use it: 1) Ensure Copilot Settings → Model Settings → Select a model with vision capabilities enabled (marked with vision icon), 2) Ensure Copilot Settings → Advanced Settings → 'Images in markdown' is turned on, 3) Set a longer context window for your selected models. You can drag images into chat or click the image button to add them as context. Copilot can analyze wireframes, diagrams, screenshots, and photos.
How do I set up my API key for Copilot?
For free users: 1) Get your API key from your provider (e.g., OpenAI Platform dashboard), 2) In Obsidian, go to Settings → Copilot Settings → Basic Settings → API Keys → Set Keys, 3) Choose your provider (OpenRouter recommended) and paste the key, 4) Select a model like gpt-4o or gpt-4o-mini, 5) Click Save and Reload. For Copilot Plus: Obtain your license key from the dashboard at obsidiancopilot.com/en/dashboard and paste it in Settings → Copilot Settings → Basic Settings → Plus License field.
What is the difference between Chat Mode and Vault QA Mode?
Chat Mode lets you reference specific notes using [[Note Name]] syntax and discuss ideas with Copilot, adding context with @. It's for focused discussions on selected notes. Vault QA Mode chats with your entire vault, retrieving relevant notes automatically using lexical and semantic search. Use Chat Mode for targeted note discussions and Vault QA for broad questions across your knowledge base like 'What are the recurring themes in my research?' Vault QA uses embeddings for semantic recall, while Chat Mode relies on explicit note references.