Frequently Asked Questions
Everything you need to know about ScholarOS. Can't find your question? Ask on GitHub.
Pricing & Plans
How much does ScholarOS cost?
ScholarOS is $10/month. This covers full access to all AI models (GPT-5, Claude 4, Gemini), unlimited chat and agent runs, knowledge graph, revision tools, PDF processing, and browser automation. No usage caps, no hidden fees.
Is there a free trial?
You can download and install ScholarOS for free. The desktop app and vault structure work without an account. AI features require an active subscription.
Can I cancel anytime?
Yes. Cancel from your dashboard at any time. Your subscription stays active until the end of the current billing period. Your vault and notes remain on your machine regardless.
Do I need to bring my own API keys?
No. The subscription covers all model access. ScholarOS proxies requests to GPT-5, Claude 4, and Gemini through OpenRouter. No separate API keys or accounts needed.
Privacy & Data
Where is my data stored?
Your vault lives on your machine as a regular folder of markdown and PDF files. Nothing is uploaded to the cloud unless you choose to sync. ScholarOS is local-first by design.
Does ScholarOS train on my notes?
No. ScholarOS never uses your notes, past papers, or any vault content to train AI models. Your data stays yours. The AI models (GPT-5, Claude 4, Gemini) are third-party services — their data policies apply to queries, but ScholarOS does not share your vault contents for training.
Is ScholarOS open source?
Yes. ScholarOS is MIT licensed. The code is public on GitHub. You can audit it, contribute to it, fork it, or run it yourself.
Does the AI search the internet?
No. ScholarOS never searches the open web by default. Your lecture notes, textbooks, and past papers define the boundary of what the AI knows. Every answer is grounded in sources you've provided.
Features
What file formats are supported?
ScholarOS supports markdown (.md), PDF (.pdf), and plain text (.txt). Obsidian vaults, Word documents, and annotated PDFs are planned for future releases.
How is this different from using ChatGPT or Claude directly?
Generic AI tools search the internet and have no memory of your specific material. ScholarOS is scoped to your vault — it never reaches beyond your notes. Every answer cites specific sources. The knowledge compounds over time instead of resetting with each conversation.
How is this different from Obsidian?
Obsidian is an excellent note-taking app. ScholarOS adds an AI layer that reads, links, and maintains a living wiki from your notes. Think of it as Obsidian + Karpathy's LLM Wiki pattern. You still own your markdown files.
How is this different from RAG (Retrieval-Augmented Generation)?
RAG re-retrieves from scratch on every query — knowledge doesn't compound. ScholarOS pre-compiles your material into structured wiki pages. Synthesis happens once, then gets updated instead of recomputed. No vector databases, no chunking, no re-ranking.
Can I use my own AI models?
Currently, ScholarOS supports GPT-5, Claude 4, and Gemini through the subscription. Bring-your-own-key support is planned for a future release.
Technical
What platforms are supported?
ScholarOS runs on macOS 12+, Windows 10+, and Linux (Ubuntu 20.04+). The desktop app is built with Electron and runs offline.
How much disk space do I need?
The app itself uses about 200 MB. Your vault size depends on your material — a typical semester's worth of notes and PDFs uses 50–200 MB. Wiki pages and outputs add minimal overhead.
Does ScholarOS work offline?
The vault and note editing work fully offline. AI features (chat, wiki compilation, revision generation) require an internet connection to reach the AI models.
Can I sync my vault across devices?
ScholarOS is local-first, but your vault is just a folder of files. You can sync it with any file sync tool (iCloud, Dropbox, Git, Syncthing). There's no built-in sync — we believe you should choose your own sync method.
Still have questions?
Ask on GitHub Discussions or check the documentation.