Free for your team. Pay only for machine work.

Machine work: external API-key requests, imports, workflows, agents and scheduled jobs. Successful external API-key reads and writes use the operation rates below. Starting a managed execution from the interface still starts that metered service.

The smallest credit pack is $5. The other one-time packs are $20 and $100. Purchased credits never expire; there is no recurring subscription or monthly minimum. Estimate usage from your team’s expected imports, integrations and workflow activity.

Each workflow run includes the first 5 CPU-sec. Code-node compute beyond that allowance uses the separate CPU-sec rate. Failed API results do not settle an App Builder operation charge; measured billable usage from a managed execution can still settle after cancellation or an error.

Each billing account includes 100,000 retained business records and 100 MB of managed file retention. Beyond those allowances, records cost $1 per 100,000 record-months and managed files cost $2 per logical GB-month. Retention is settled daily. Archived and Trash records still count. Shared file references count once; independent copies and saved versions count separately.

Ordinary manual work remains available at a zero balance. New metered work pauses until the billing account has enough credits. Model usage and the Proto Agent fee are described below; they are additional to the operation rates in this table. The platform pricing page carries the current full rate card and account rules.

Fitting the system to credit officers, independent credit approvers and lending operations teams at banks and commercial lenders, migrating data and going live by process is priced separately. Your quote reflects: sites, integrations, data volume and support level. Talk to our team about your implementation plan.

Use Proto to configure and later change the application on a branch you review before release. Its machine work and model usage are metered.

ERP-1 Fast costs $0.05 per 1 million cached input tokens, $0.25 per 1 million uncached input tokens and $0.75 per 1 million output tokens. The 8% Proto Agent fee is added to the exact model amount and shown separately on the receipt.

With an open router key, open router bills model inference directly. ERP.AI charges 8% of the exact open router-reported model cost, plus any separately metered ERP.AI resources used by the run. Usage receipts identify the model, credential source, applied rate-card version and itemized charges.