Free to start. Priced by memory, not seats.
Prototype for free, pay a little when you ship, and talk to us when a fleet depends on it. Self-hosting the open-source core is always free.
Free
Free foreverPerfect for prototyping, local embedded engines, and individual agent or robot nodes.
Start Building Free- Up to 3 memory databases & 1 workspace
- 10,000 write operations/mo
- 100,000 vector memory entries
- Fast hybrid vector & keyword retrieval
- Native MCP protocol & ROS2 bridge access
- Open-source SDKs (Python, TS, ROS2)
- Community support & Discord
Starter
For autonomous agents and robots running in production with persistent state.
Launch Agent Memory- 100,000 write operations/mo
- 1,000,000 vector memory entries
- 5 memory databases across 3 workspaces
- Zero-copy state branching
- Automatic entity deduplication
- Email support
Pro
Most popularFor multi-agent systems, robot fleets, and production cognitive memory workloads.
Start Operating at Scale- 1,000,000 write operations/mo
- 10,000,000 vector memory entries
- 25 memory databases across 10 workspaces
- Full Knowledge Graph & entity linking
- Zero-copy branching & state merging
- Automated reflection & insight distillation
- Priority support & Slack access
Enterprise
Dedicated isolated clusters, robot fleet synchronization, and custom SLAs.
Contact Fleet Sales- Unlimited vector & graph memory
- Dedicated isolated cloud clusters
- Fleet robotics streaming synchronization
- SAML SSO, RBAC & custom audit logging
- VPC peering & on-premise options
- Dedicated solutions engineer & custom SLA
Which plan fits?
Slide to your expected size. Limits are per workspace.
$12 a month, or less billed yearly.
Every plan, line by line.
| Feature | Free | Starter | Pro | Enterprise |
|---|---|---|---|---|
| Usage | ||||
| Vector memories | 100k | 1M | 10M | Unlimited |
| Writes / month | 10k | 100k | 1M | Custom |
| Memory databases | 3 | 5 | 25 | Unlimited |
| Workspaces | 1 | 3 | 10 | Unlimited |
| Memory | ||||
| Hybrid vector + keyword retrieval | ||||
| Hosted MCP server | ||||
| Zero-copy branching | ||||
| Merge strategies | ||||
| Knowledge graph + entity linking | ||||
| Reflection and contradiction detection | ||||
| Importers (Mem0, Zep, Pinecone…) | ||||
| Team and security | ||||
| Scoped API keys | ||||
| Roles and invitations | ||||
| SAML SSO | ||||
| Dedicated infrastructure, VPC | ||||
| Support | ||||
| Support | Community | Priority + Slack | Dedicated engineer, SLA | |
Recommend Axonio, earn 25% for the life of the customer.
Billing questions.
Is the Free tier really free forever?
Yes. Three memory databases, 100,000 vector memories and full MCP access, with no credit card and no expiry.
What counts as a write?
Any operation that stores or changes memory: remember, update, delete, or a memory added inside a transaction. Retrievals are metered separately against generous per-plan limits.
What happens if I go over my limits?
New writes are rejected with a clear quota error until you upgrade or your monthly usage resets. Everything already stored stays in place, and current usage is always visible in the dashboard.
Can I self-host instead of paying?
Yes. The core engine is MIT-licensed and free to self-host with no limits. Paid plans are for the managed cloud, hosted MCP and team features.
How does annual billing work?
Pay yearly and get two months free: Starter is $10 a month and Pro is $39 a month, billed annually.
Do you have an affiliate program?
Yes. Affiliates earn 25% of the revenue from customers they refer, every month, for as long as the customer stays.