FAQ
Moss — Frequently Asked Questions
What is Moss?
Moss is a personal AI with memory that carries context across every conversation — accurate recall at 7M+ tokens, where Claude degrades past 1M. Context is surfaced before you ask, not fetched on demand. See the benchmark.
How is Moss different from Claude or ChatGPT's native memory?
Claude and ChatGPT fetch a shallow facts table on demand and degrade once the relevant context lives in a different thread or past roughly 1M tokens. Moss carries context across every conversation, stays accurate at 7M+ tokens, and surfaces relevant past conversations before you ask.
Do I pick which AI model to use?
No. Moss orchestrates Claude, GPT, Gemini, Groq, and Perplexity server-side. Your tier unlocks an arsenal of models; Moss routes each memory operation — retrieval, abstraction, synthesis — to the right one. You just ask and get the best answer.
Can I import my existing ChatGPT, Claude, or Gemini conversation history?
Yes. Upload your export file on the Sprout tier ($25/mo) or above. Moss extracts semantic units from every conversation, indexes them permanently, and makes them searchable across every future conversation.
What does Moss cost?
Free (chat-only, 15 exchanges per day, $0/mo). Seed ($9/mo, every-day memory without daily caps). Sprout ($25/mo, document uploads + research). Grove ($49/mo, premium models + verification). Thicket ($99/mo, priority access + extended memory). Canopy ($249/mo, enterprise-only by enquiry).
Where does my data live?
Your data is scoped to your account with row-level security on every table. You can export your memory on demand and delete facts or whole conversations at any time. Moss never trains on user data.
Try Moss · Home · Benchmark · Blog · Privacy