open-notebook

Open Notebook is an open-source, self-hosted research and knowledge tool positioned as a privacy-focused alternative to Google's Notebook LM. It is MIT licensed and built with Python, FastAPI, Next.js, React, and SurrealDB. Core capabilities described in the README: - Multi-notebook organization for managing separate research projects. - Content ingestion for PDFs, videos, audio, web pages, Office documents, and other formats. - Multi-model AI support through the Esperanto library, with a provider matrix covering OpenAI, Anthropic, Groq, Google GenAI, Vertex AI, Ollama, oMLX, Perplexity, ElevenLabs, Deepgram, Azure OpenAI, Mistral, DeepSeek, Cohere, Voyage, xAI, OpenRouter, DashScope (Qwen), MiniMax, Novita, PayPerQ, and OpenAI-compatible endpoints such as LM Studio. Providers vary in LLM, embedding, speech-to-text, and text-to-speech support. - Full-text and vector search across stored content. - Context-aware chat over research materials, with multiple chat sessions per notebook. - AI-assisted and manual note creation. - Content transformations for summarizing and extracting insights. - Podcast generation with 1-4 speakers and configurable Episode Profiles. - Reasoning model support, including DeepSeek-R1 and Qwen3. - A REST API for programmatic access, plus MCP integration for clients such as Claude Desktop and VS Code. - Optional password protection for public deployments. - Multi-language UI: English, Portuguese, Simplified and Traditional Chinese, Japanese, Russian, and Bengali. Deployment is primarily via Docker Compose, with a SurrealDB service and an application service exposing a web UI on port 8502 and a REST API on port 5055. The README notes that database credentials default to root:root for local use and should be overridden before network exposure, and that an encryption key must be set to protect stored API keys. Alternative installation paths include running with Ollama for local models and building from source for development. The README also outlines a roadmap including live front-end updates, asynchronous processing, cross-notebook sources, and bookmark integration, and lists recently completed work such as the Next.js frontend, REST API, multi-model support, podcast generation, content transformations, enhanced citations, and multiple chat sessions. Community support is available through Discord, GitHub Discussions, and GitHub Issues.