Build something that works, on the first paid plan.

Each successful metered API response counts as one reading, with returned rows and response bytes tracked separately. Charts, prediction-price history and historical signal replay are metered. Current tool tables, discovery, account reads and the research liquidation map do not consume paid readings. Per-minute workspace limits apply across the dashboard, REST and MCP.

Usage is metered per month and per minute. Past the limit the API returns a clear error rather than quietly serving stale readings or adding to your invoice, so a runaway script cannot increase what you pay. Planned, not yet live: warnings at 70% and 90% and when the current rate would empty the allowance early, plus extra readings in blocks of 100,000 for $19 — off by default, with a monthly ceiling you set.

Not yet. There is a free workspace instead, with no card: the documentation, API keys and the schema. A 14-day trial of the full Pro plan, started deliberately rather than at sign-up, is planned; when it ends the account returns to Free and any integration is told plainly that its access changed.

No. Annual billing is one payment, but readings and snapshots renew every calendar month exactly as they do on monthly billing.

Through Paddle, which acts as the merchant of record. Choose monthly or annual at checkout. Prices on this page are in USD and exclude tax; your currency and local VAT or sales tax are shown at checkout before anything is charged.

Yes. Subscriptions renew for the period you selected and can be cancelled before renewal. Access runs to the end of the period you have already paid for.

Both exist and they are labelled differently. Every model states when its history begins, which instruments it covers, and whether a reading was actually published at that moment or reconstructed afterwards from historical data. That distinction is the same on every plan. A longer window can be sold; an honest description of it cannot be withheld from the cheaper one.

Leverage Stress, Flow Absorption and Positioning Divergence. Cascade Risk, Expectations Gap and Trader Conviction need account-level history and cross-market matching before they are worth shipping, and are marked as in development wherever they appear. The library section above says which is which.

On Pro and Scale: use them in your own strategies, bots and internal work, including as a business. Embedding readings in a product you sell to others, redistributing them or white-labelling them is a separate conversation about scope and rights.

All six core models are included, and models of the same kind extend that library. Expensive satellite models, dedicated regions or bespoke computation may later carry their own module or add-on rather than being folded in silently.

No. KnownAt is data infrastructure. We do not recommend a security, market, strategy or transaction, and research conclusions and trading decisions remain yours.