Simple starting point
Start with memory primitives. Scale into a managed AI backend.
Use the plan language as a scaffold until commercial pricing is finalized. The structure mirrors common BaaS expectations: free development, usage-based growth, and enterprise controls.
- Billing model
- Project plan plus compute, storage, and AI usage
- Deployment
- Hosted cloud, private cloud, or self-managed
- Enterprise
- Security review, governance, custom support
Developer
For prototypes, research apps, and local evaluation.
$0
- One hosted project
- PostgreSQL Memory Core
- Starter ingestion and hybrid search
- Community support
Team
For production AI apps and internal agent workflows.
Usage
- Multiple projects and environments
- Active Memory Pools
- MC2DB tool governance
- Usage-based compute, storage, and model calls
Enterprise
For regulated memory, private deployment, and platform teams.
Custom
- Private networking and deployment options
- Advanced audit and access controls
- Security and architecture review
- Dedicated support path
What scales
Pay for the infrastructure your memory uses.
The pricing surface is intentionally familiar for teams coming from BaaS platforms: database compute, source storage, vector indexes, function execution, bandwidth, and model usage.
| Meter | What it covers | Why it matters |
|---|---|---|
| Database compute | PostgreSQL CPU and memory | Runs application data, memory catalog, policies, and SQL workflows. |
| Source storage | Documents, transcripts, logs, and raw source files | Keeps evidence available for reprocessing and audit. |
| Hybrid retrieval | Embeddings, graph edges, and query execution | Combines semantic recall with relational, graph, and temporal constraints. |
| Functions | Ingestion, extraction, validation, and rebuild jobs | Runs memory workflows without a separate backend service. |
| Model usage | LLM and embedding provider calls | Supports extraction, summarization, classification, and agent workflows. |