Estimated Cost

Strata can put a rough monthly price tag on your design so you can compare options and spot the expensive pieces while you’re still drawing — long before anything is deployed.

Turn it on

Click the Cost toggle in the top toolbar. When it’s on, the button shows the estimated monthly total for the whole diagram (e.g. Cost ~$184/mo) and each priced node gets a small cost badge on the canvas. Click it again to turn the estimate off.

What the estimate covers

The figure is a back-of-the-envelope heuristic, not a billing forecast. Each resource is priced from its service type, with light refinement from its configuration for the big-ticket services, including:

• Instance size / class — larger compute and database instances cost more.
• Count / capacity — desired capacity, node count, replica count, and similar multipliers scale the estimate.
• Multi-AZ — highly-available databases roughly double.
• Storage — allocated storage and disk size (GiB) are factored in.

Structural, free resources (VPC, subnet, route table, security group, …) count as $0. Services Strata can’t price yet show — and don’t contribute to the total.

The total is a floor, not a bill

Unpriced resources fall into two classes, and Strata keeps them honest:

• Free / control-plane services (IAM, ACM, CloudFormation, Organizations, …) legitimately cost ~$0 and don’t affect the total.
• Billable-but-unmapped services are real spend Strata can’t price yet. When any are present, the total is marked as a FLOOR (a lower bound) and the unmapped types are listed — so a redundant serverless minimum repeated across many accounts can never masquerade as $0.

The MCP estimate_cost tool and the strata cost CLI both surface the floor flag and the unmapped types explicitly. See Multi-Account Cost Model for the offline, all-environments workflow.

By default the estimate ignores data transfer, request volume, commitment discounts, and regional price differences. The estimate_cost tool and strata cost accept assumptions — regionMultiplier, hoursPerMonth, and discountPct (Savings Plan / RI) — to bring the figure closer to a real bill. Treat it as a way to compare designs and find cost hotspots — not a quote.

Deeper cost optimization

Beyond this heuristic, region-aware usage-based pricing, rightsizing and idle recommendations, and commitment-discount awareness are on the roadmap. The cost-optimization prompt in the app lets you register interest so we can gauge demand before building it — see the roadmap for where this is headed.