> ## Documentation Index > Fetch the complete documentation index at: https://docs.rippit.com/llms.txt > Use this file to discover all available pages before exploring further. # Billing and Credits > How Rippit charges for work: Credits, free grants, purchases, auto-reload, platform costs, provider credits, and how to estimate a cost before you spend. Rippit charges for the work it does, from one prepaid balance called Credits. This page explains what spends that balance, what is billed separately, and how to find a cost before and after you run something. ## The short version * **Credits are a prepaid balance in US dollars.** A new workspace gets some free, and a workspace admin buys more. * **AI work spends Credits as it runs.** The cost is the model tokens used, plus any search or other tools the work calls. * **Storage is a monthly platform cost.** It is taken from Credits first, and only a shortfall is invoiced. * **Provider credits are a separate unit.** Some lookups report credits spent with a search or contact provider, on top of the Credits spent on the model. * **When Credits run out, AI work stops.** In chat, Rippit pauses, and you send the message again once Credits are added. A [spending limit](#spending-limits) stops work the same way. Billing actions and billing details need a workspace admin. ## Two kinds of account | | Self-serve | Contract | | - | - | - | | How you pay | Prepaid Credits, bought by card | An AI credit budget agreed with Rippit | | Adding more | Buy Credits or turn on auto-reload | Through your Customer Success Manager | | Platform costs | Monthly, from Credits | Not billed this way | | What billing details show | Balance, free grants, spend, payment method, auto-reload, payment issues | Budget, spend, remaining, and spend by feature | The rest of this page describes self-serve workspaces unless it says otherwise. ## What spends Credits | Work | What it costs | | - | - | | Enrichments in a worksheet or pipeline | One model call per row. Columns defined together share that call, so ten columns cost about the same as one | | Asking Rippit in chat | The model tokens and tools each answer uses | | Agent runs in a pipeline | One agent session per row, plus whatever the agent reads with its tools | | [External research](https://docs.rippit.com/pipelines/research.md#cost) | The model doing the research, plus the search provider's cost | | Pipeline tests | The same as the real work, on the sampled rows | | Storage | A monthly platform cost. See [Platform costs](#platform-costs) | Formulas cost nothing, rollups cost almost nothing, and rows excluded by a filter are never processed. A live pipeline keeps spending on its schedule until you pause it. ## Free Credits | Grant | Amount | When | | - | - | - | | Starting balance | \$10 | When the workspace is created, including a [temporary account](https://docs.rippit.com/temporary-accounts.md) | | Verified card | \$90 | Once per workspace, when a workspace admin saves and verifies a billing card | Replacing a card later does not grant more. The balance is free grants plus purchases, less AI spend and platform costs. ## Buying Credits A workspace admin buys Credits on a Stripe checkout page or by charging the saved card. A saved-card purchase is a whole-dollar amount from $1 to $10,000, plus tax, and it charges immediately. A checkout link charges nothing until someone pays on it. Credits are added when the payment completes, and can be spent right away. The balance can take a moment to update. An assistant or script should charge a saved card only when a person asked for that exact purchase. If the result of a purchase is unknown, check its status before buying again. Checking never charges. ## Auto-reload Auto-reload buys Credits on the saved card when the balance falls to a trigger amount. It is off until a workspace admin turns it on. | Setting | Default | Notes | | - | - | - | | Trigger balance | \$20 | Whole dollars | | Top up to | \$100 | At most $10,000, and at least $1 above the trigger | | Monthly limit | \$500 | Before tax, from $1 to $10,000, per UTC calendar month | Once a month's reloads reach the monthly limit, auto-reload stays on but buys nothing more until the next month. After repeated failed payments it switches off. A top-up that needs confirming with your bank waits on the Billing page. ## Spending limits Separately from the balance, a workspace has limits on AI spend over shorter windows, such as a day. When one is reached, AI work is refused even if Credits remain, and can run again once that window has passed. A refusal for a spending limit and a refusal for an empty balance look the same to an assistant, so check the balance to tell them apart. ## Platform costs Rippit records the workspace's storage every day and charges for it once per billing period. The period is a month that starts on the day the account was created. The charge is deducted from available Credits. If the balance does not cover it, the rest is invoiced to the saved card. An unpaid platform invoice appears as a payment issue with a link to resolve it. Platform costs are reported separately from AI usage, so the spend figures for today, this week, and this month cover AI work only. ## Provider credits Credits pay for Rippit's work. Provider credits are a different unit, counted by the search or contact provider a lookup used. * **Contact lookups in research** (work email and phone) use provider credits on top of the Credits for the model. They run only when you ask for them, and Rippit reports the credits each one used. * **Connected MCP servers** such as [Clay](https://docs.rippit.com/agentic-chat/connectors/clay.md) spend your own account with that vendor, under the limits you set there. Rippit Credits still pay for the model doing the work. An unknown provider charge is reported as unknown, never as zero. ## Estimate a cost before you spend | What you want to know | How | | - | - | | What a model costs | `list_ai_models` or `rippit data models` lists each model's price per million input and output tokens | | What a pipeline will cost | Run a test, then ask for the estimate: `get_pipeline_cost_estimate` or `rippit pipelines estimate PIPELINE_ID` | | What a worksheet enrichment will cost | Enrich a sample of rows first, then scale by row count | | What research or an agent run will cost | Test on a few rows and check what the run spent | The pipeline estimate gives the one-time cost of the rows you already have and the ongoing cost per day. It is built from the latest finished test, so it is unavailable until one finishes. It counts the model calls that fill enrichments, not agent runs or search providers, so a pipeline that uses those costs more than the estimate shows. There is no estimate for a worksheet enrichment before it runs. Cost scales with the number of rows, not the number of columns, so a sample gives a reliable per-row figure. ## See what was spent | Where | What it shows | | - | - | | **Billing** page in the app | Credits, AI usage over time, platform usage, payment method, auto-reload, payment issues | | `get_billing_usage` | The same details for an assistant, plus the table IDs of the usage logs | | `rippit billing usage` | The same details in a terminal. Add `--format json` for everything | | Usage logs | The AI usage log, storage snapshots, and platform billing history are tables. Read them with `describe_table` and `read_table` | | A pipeline's canvas | What the pipeline's live runs spent over the last 30 days, or what one test spent, with a link to those rows in the AI usage log. Workspace admins only | | `rippit pipelines get PIPELINE_ID` | The AI cost of one pipeline's runs over the last 180 days | A null or missing field means the figure was unavailable, not that nothing was spent. See the [CLI billing commands](https://docs.rippit.com/cli.md#billing-and-credits) and the [billing tools](https://docs.rippit.com/mcp-server/tools.md) an assistant can use. This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.