Plans & Pricing
Start free, scale as you grow. Every plan includes TurboQuant compression and semantic search — upgrade for advanced models, team collaboration, and enterprise control.
Free
Get started with AI agent memory — no credit card required.
- 1,000 memories
- Basic embedding model
- Single project
- REST API access
- TurboQuant compression
- Community support
Pro
Unlock advanced models, higher limits, and priority processing for production workloads.
- 50,000 memories
- Advanced models (BGE-M3, 1024-dim)
- Unlimited projects
- Priority embedding queue
- TypeScript & Python SDKs
- Version history
- Email support
Team
Collaborate on memory projects, manage teams, and gain full infrastructure visibility.
- Unlimited memories
- Organization management
- Team collaboration
- Role-based access control
- Usage analytics & reporting
- MCP server access
- Priority support
Enterprise
For organizations that need full control, compliance, and dedicated support.
- On-prem deployment
- Bring your own embeddings (BYOE)
- Custom compression configurations
- SLA & uptime guarantees
- Dedicated account management
- Audit logging
- Custom integrations
Detailed Comparison
See exactly what's included in each plan
| Feature | Free | Pro | Team | Enterprise |
|---|---|---|---|---|
| Memory storage | 1,000 | 50,000 | Unlimited | Unlimited |
| Projects | 1 | Unlimited | Unlimited | Unlimited |
| TurboQuant compression | ||||
| Semantic search | ||||
| REST API | ||||
| Version history | ||||
| Basic embedding model (768-dim) | ||||
| Advanced models (BGE-M3, 1024-dim) | ||||
| Priority embedding queue | ||||
| TypeScript SDK | ||||
| Python SDK | ||||
| MCP server access | ||||
| Organizations | ||||
| Team collaboration | ||||
| Role-based access control | ||||
| Usage analytics & reporting | ||||
| On-prem deployment | ||||
| Bring your own embeddings (BYOE) | ||||
| Custom compression configurations | ||||
| Audit logging | ||||
| SLA & uptime guarantees |
Frequently Asked Questions
Can I switch plans at any time?
Yes. Upgrade, downgrade, or cancel at any time. When upgrading, you get immediate access to new features and higher limits. When downgrading, your current billing cycle completes before the change takes effect.
What counts as a 'memory'?
Each piece of text content stored via the API or SDK counts as one memory. Updating an existing memory creates a new version but does not count as an additional memory against your limit.
Do I need a credit card for the Free plan?
No. The Free plan requires no credit card and no payment information. Just sign up and start storing memories.
What embedding models are available?
The Free plan uses the basic nomic-embed-text model (768 dimensions). Pro and above unlock BGE-M3 (1024 dimensions) for higher quality semantic search. Enterprise customers can bring their own embedding models.
How does TurboQuant compression work?
TurboQuant uses a two-stage pipeline: PolarQuant applies Lloyd-Max quantization to reduce vector precision to 3 bits, then QJL residual correction preserves cosine similarity accuracy. The result is a 95% reduction in vector storage with minimal search quality loss.
Can I self-host MemTurbo?
Yes. MemTurbo is open-source and can be self-hosted with Docker Compose (PostgreSQL + Redis). Enterprise customers get dedicated support for on-prem deployments, custom configurations, and SLA guarantees.
What is MCP server access?
MCP (Model Context Protocol) allows AI assistants like Claude to directly access MemTurbo memories as tools. This enables agents to store and retrieve memories natively without custom API integration code.