Pricing
Free where it runs.
Paid where it syncs.
The engine is open source and free forever on your machine. You pay when memory needs to travel — across your machines, or across your team.
Local
The full engine, on your machine.
- ✓Unlimited sessions indexed
- ✓Claude Code, Codex & Cursor
- ✓Anchored section retrieval
- ✓Semantic search + reranking
- ✓All agent skills & MCP server
- ✓Open source, no account needed
curl -fsSL https://tryarca.ai/install | bashPro
Most popularYour memory, on every machine you use.
- ✓Everything in Local
- ✓Encrypted sync across machines
- ✓Hosted backup of your corpus
- ✓Extended archive index (all history)
- ✓Web viewer for sessions & sections
- ✓Priority support
Team
One memory across every engineer's agents.
- ✓Everything in Pro
- ✓Shared team memory with attribution
- ✓Section-scoped sharing policies
- ✓Repo & project-level access control
- ✓Team-wide search from any agent
- ✓SSO & admin controls
Enterprise
Self-hosted sync, air-gapped deployments, custom retention, SLAs.
FAQ
Pricing questions.
What am I actually paying for?+
Movement, not the engine. Indexing, retrieval, agent skills, and the MCP server are free and open source forever. Paid plans add hosted sync — your memory following you across machines (Pro) or shared across a team with attribution and access control (Team).
Can I self-host the sync layer?+
Enterprise plans include self-hosted and air-gapped deployment of the sync server. For Pro and Team we run it for you, encrypted in transit and at rest.
Do you train models on my sessions?+
No. Never. Your sessions are your engineering history, not our training data.
What happens if I cancel?+
Your local memory keeps working — it never depended on us. Synced data is exportable as plain markdown and JSON before your account closes.
Is there a free trial for paid plans?+
Yes, 14 days on Pro and Team, no credit card up front. And the Local tier is a permanent free trial of everything that matters.