Pick a tier, pay monthly, use it. Your license activates immediately and renews on the same day next month. Cancel anytime from your dashboard — no prorated refunds, but you keep full tier access through the end of your billing cycle. If you don't renew, your proxy drops back to the free developer key features. Nothing breaks.

Developer (Free): persistent memory, basic knowledge graph, local agent mode, local image gen, 5-user isolation, 5K facts, read-only cloud OAuth. Standard ($5/mo): Ghost Agent, memory consolidation, embedding re-ranking, persona learning, conversation branching, brain dashboard, remote HTTPS access, parental controls, encrypted backups, TrimScript plugins, 25K facts. Pro ($15/mo): full hybrid cloud+local routing, Web Scout, sub-agent swarm, hybrid workflow planning, cost tracker, model aliases, context selector, 100K facts. Premium ($30/mo): visual knowledge graph explorer, n8n/MCP integration, Claude Code subprocess, Ghost Mesh (multi-agent collab), ambient task tracking, memory staleness transparency, unlimited everything, SLA support.

This is why Pro exists. Your local LLM is fast and free but weaker on hard problems; Claude and GPT are expensive but precise. Pro's hybrid router keeps them working hand-in-hand: you draft and iterate with your local model, and when you hit something tricky — a gnarly refactor, an unfamiliar API, a performance question — the request quietly escalates to the cloud. Both sides see the same memory (your project history, preferences, past decisions), so the cloud response picks up mid-thought. Commit messages locally, code review in Claude, both feeding the same knowledge graph.

You can, and Developer already gives you read-only OAuth for one provider. What Pro adds is the orchestration: smart routing decides which model for which request, cost tracking per-call, your persistent memory automatically injected into both sides, prompt enhancement that makes short asks worth 2-3x more, and conversation history you can grep months later. Raw API access is a chatbox; Pro makes it a personal AI that knows you.

v1.0 is live on GitHub today with the free developer key. v2.0 is in active development — a much bigger release with hybrid cloud routing, the full intelligence stack, visual knowledge graph, and all the features described in Pro and Premium. Early adopters who subscribe to any paid tier at launch get every v2.0 feature rolled out as it ships, no price change. Watch the GitHub repo for the v2.0 release notes.

Yes. Use the contact page for volume discounts, custom deployment, on-prem licensing, or onboarding help. Teams of 10+ get discounted seats.

That's fine — and it's fully functional. The free developer key on GitHub gives you persistent memory, context trimming, up to 5 users, file uploads (15 MB), and the dashboard. No sign-up, no credit card, no expiration. Paid tiers just add the extras when you want them.

Only the specific request you route there — and only if you've opted in via OAuth. Your memory, files, history, and identity all stay on your machine. Hybrid mode sends the minimum prompt to the cloud provider, gets the response back, and stores nothing with them. You can inspect every request in the brain dashboard before it goes out.