Plans#

Parametric Memory offers four paid tiers. Every customer gets a dedicated instance with real infrastructure — there is no unpaid access level.

The tiers#

What "bootstraps" means#

A bootstrap is one call to memory_session_bootstrap — the call your AI makes at the start of each session to load relevant context. Each session typically uses one bootstrap. 200/month covers roughly 6 Claude sessions per day; 1,000/month covers roughly 33.

What "atoms" means#

An atom is one named piece of memory — one key/value pair. 1,000 atoms covers approximately 2 months of daily use. 10,000 atoms covers approximately 18 months of daily professional use for a solo developer. You can check your current usage on the Dashboard at any time.

Overage pricing#

If you exceed your atom limit, atoms are charged at $0.001 per atom above your tier limit. Overages are reported to Stripe via metered billing and appear on your next invoice. You can set a spend cap to prevent unexpected overage charges.

Starter vs Solo — which should I choose?#

• Starter if you are trying Parametric Memory for the first time, or use Claude occasionally
• Solo if you use Claude for personal projects, solo development, or moderate daily AI use

Solo vs Professional — which should I choose?#

• Solo if you have a single main project and moderate daily AI use
• Professional if you run multiple agents, have intensive daily use, or frequently hit the bootstrap limit on Solo

When to choose Team#

Team gives you a dedicated DigitalOcean droplet rather than a shared container. Choose Team when:

• You need guaranteed compute isolation from other customers — your instance runs on its own server with no shared resources
• You want 500,000+ atoms for large-scale institutional memory
• Multiple people on your team will share the same memory instance
• You need multiple memory substrates for different projects or teams (multi-substrate is rolling out to Team customers — contact us)
• You want the best possible response times for bootstraps and large batch operations

What "dedicated" means in practice: On shared tiers, your container runs alongside other customers on the same server. Resources are allocated per container, but you share the underlying host. On Team, you get an entire server to yourself — provisioned from a pre-built image for faster startup, running in your chosen region, with full CPU and memory exclusively for your workloads.

Money-back guarantee#

All paid plans include a 30-day money-back guarantee. If Parametric Memory does not work for your use case, contact support within 30 days of your first payment for a full refund.

Changing plans#

You can upgrade or downgrade at any time from the Dashboard. See Upgrade Plan and Downgrade Plan for the full details of what happens to your instance during each transition.