How do I set up Ollama to use local LLMs with ChatGPT MD?

Install Ollama from ollama.ai, pull a model like llama3.2 via terminal, then configure the Ollama URL and default model in ChatGPT MD settings. You can override models per note using frontmatter with the 'ollama@' prefix.

ChatGPT MD vs Obsidian Copilot: which is better for AI note-taking?

ChatGPT MD excels in privacy and multi-provider support with local LLM options, while Copilot might offer tighter Obsidian integration for specific tasks. Choose ChatGPT MD if you need offline capabilities or access to models like Claude via OpenRouter.

Can I use ChatGPT MD to search the web for real-time information?

Yes, by enabling tool calling and adding a Brave Search API key, you can use models like GPT-4o-search-preview or Perplexity's online models via OpenRouter.ai for web searches with human approval for each query.

How to create a custom AI agent for coding assistance in ChatGPT MD?

Use the AI Wizard or create a Markdown file with frontmatter defining model, temperature, and a system prompt in the body. Then apply it to notes using the 'Choose Agent' command for reusable personas.

Is my data safe when using cloud AI services with ChatGPT MD?

Data sent to cloud APIs depends on the provider's policies, but tool calling for vault operations keeps data local. For full privacy, use local LLMs via Ollama to avoid any data leaving your computer.

How to change the AI model for a specific conversation in ChatGPT MD?

Add frontmatter at the top of your note with the 'model' parameter, e.g., 'model: gpt-5' for OpenAI or 'model: ollama@llama3.2' for local LLMs, and use the chat command to apply it instantly.

Open-Awesome

ChatGPT-MD

MITTypeScript3.1.0

An Obsidian plugin that integrates ChatGPT, OpenRouter.ai, and local LLMs for AI-powered conversations directly within your notes.

GitHub

1.4k stars103 forks0 contributors

What is ChatGPT-MD?

ChatGPT MD is an Obsidian plugin that integrates AI assistants directly into your note-taking environment. It allows users to chat with various AI models—including cloud services like OpenAI and OpenRouter.ai, or local LLMs via Ollama—without leaving their vault, enabling AI-powered brainstorming, research, and content generation within their existing workflow.

Target Audience

Obsidian users who want to incorporate AI assistance into their note-taking, knowledge management, and writing processes, particularly those valuing privacy, offline capabilities, and deep integration with their vault.

Value Proposition

Developers choose ChatGPT MD for its seamless Obsidian integration, privacy-first design with local LLM support, and flexible multi-provider architecture that avoids vendor lock-in while keeping conversations contextual within their notes.

Overview

A (nearly) seamless integration of ChatGPT into Obsidian.

Use Cases

Best For

Adding AI-powered research and brainstorming capabilities to Obsidian notes
Running private, offline AI conversations using local LLMs like Llama or Gemma
Creating reusable AI personas for specific tasks (e.g., coding assistant, editor)
Enhancing note-taking with real-time web search via AI models
Building AI-augmented knowledge bases with vault-aware tool calling
Experimenting with multiple AI providers (OpenAI, Claude, Gemini) in one interface

Not Ideal For

Teams requiring fully autonomous AI agents that execute tools without manual approval
Users of alternative note-taking apps like Notion or Roam Research seeking built-in AI features
Projects needing a standalone, cross-platform AI chat interface without Obsidian dependency

Pros & Cons

Pros

Multi-Provider Flexibility

Supports OpenAI, OpenRouter.ai, and local LLMs via Ollama/LM Studio, allowing seamless switching between cloud and offline models for diverse AI needs.

Privacy-First Tool Calling

Implements human-in-the-loop approval for vault search and web queries, ensuring data stays local unless explicitly shared, as highlighted in the v3.0.0 release notes.

Deep Obsidian Integration

Enables AI conversations directly within notes, with per-note configuration via frontmatter and commands accessible from Obsidian's palette for a seamless workflow.

Local LLM Offline Support

Allows complete offline AI usage with Ollama or LM Studio, avoiding API costs and enhancing privacy, as detailed in the setup guides for local models.

Cons

Obsidian-Exclusive Dependency

Only functions within Obsidian, making it useless for users of other note-taking platforms or those wanting a standalone application outside the vault ecosystem.

Manual Approval Workflow

Tool execution requires user approval for each step, which can disrupt automated processes and slow down productivity, despite privacy benefits.

Local LLM Setup Complexity

Configuring Ollama or LM Studio involves separate installations, model downloads, and URL settings, adding overhead compared to cloud-only plugins.

Frequently Asked Questions

Related Projects

PandasAI

Chat with your database or your datalake (SQL, CSV, parquet). PandasAI makes data analysis conversational using LLMs and RAG.

Stars23,620

Forks2,333

Last commit8 months ago

chatgpt-vscode

A VSCode extension that allows you to use ChatGPT

Stars4,940

Forks361

Last commit2 years ago

AICommand

ChatGPT integration with Unity Editor

Stars4,113

Forks426