Free to start. Pay when you ask more.
Every plan includes a number of questions a month. Free stops at 200; paid plans keep answering past their allowance and charge $1 for each extra 100.
Free
$0
forever · no card
For trying it on one database and seeing the SQL that runs.
One database
200 questions a month
Ask questions in plain English, and see the SQL that ran
Choose exactly which tables and columns the model can read
Write table and column guidance to correct how it queries
Query history and per-database analytics
Your own model key: Bedrock, Azure, Ollama, vLLM, Anthropic, OpenAI or Gemini
Questions past 200 wait for next month
Teammates, invites and roles
Starter
Recommended$10
per month
For one person who asks their database questions every day.
Everything in Free
2,000 questions a month
Extra questions at $1 per 100, never a hard stop
Cancel anytime
More than one database
Teammates, invites and roles
Pro
$29
per month
For one person whose questions span more than one database.
Everything in Starter
Up to 3 databases
100,000 questions a month
Switch between them without signing out
Teammates, invites and roles
Review queue
Testing mode and version history
Small Team
$79
per month
For a team where the people asking questions are not the people who know the schema.
Everything in Pro
Up to 5 databases
100,000 questions a month, shared
Invite teammates by email
Five roles, granted per database
Review queue — triage what your team flagged as wrong
Testing mode, sandboxes, and version history
Prices are in US dollars and billed monthly through Polar, our payment provider. They cover the product, not model tokens — those stay on your own provider key. On paid plans, questions past the monthly allowance are $1 per 100, added to the next invoice. Have a promo code? Enter it when you choose a plan, or at checkout. Cancel anytime from Billing; your plan runs to the end of the month you paid for. Need SSO or unlimited use? Enterprise is coming — contact Irevex.
Bring your own key and the metered part is yours.
Every question sends a request to a language model, and that is metered by whoever provides it. An instance can carry its own API key — encrypted at rest with Fernet, never returned to the browser after it is saved — in which case you pay your provider directly and BeyondQueries charges only for the plan. It is both the cheapest way to run this and the one that keeps the prompt traffic on an account you control.
Retrieval reduces that bill over time as a side effect: a question answered from approved grounding is a shorter prompt than one that has to reason about the whole schema. How the loop works.
Two things people check first.
What happens when I run out of questions? On Free, new questions wait until the count resets on the 1st of the month, or you upgrade. On Starter, Pro and Small Team you keep asking: questions past the allowance cost $1 per 100. Everything asked on a team's databases counts toward that team's allowance.
What happens to our corrections if we downgrade? They stay. The knowledge base belongs to the instance, and dropping to Free stops new approvals rather than deleting past ones. Export is not yet one click — ask and it will be provided.
Anything else is probably in the FAQ, including exactly which role can see whose queries.
Start on Free. No card, so the only cost is your time.
Connect a database with a read-only user, ask three real questions, and correct the first thing it gets wrong.