Billing & credits

Quanty bills through Stripe subscriptions, and meters AI work in credits. Your plan includes a monthly credit allowance; one-time top-up packs extend it when you need more. Everything AI does, from chat replies to document extraction, draws from the same balance, so there is a single number to watch.

What a credit is

A credit is our billing unit for AI work — not the same thing as a language model's token. One Quanty credit is $0.001 of AI usage at our published rates. A published rate is the model provider's list price times 1.3 — the same multiple for every model, with no exceptions.

Rates are per million model tokens, split between input (what the model reads) and output (what it writes). Every model in the table is served through Amazon Bedrock (AWS).

Claude 5 models become selectable in a workspace as soon as AWS enables them for it; until then a request for one of them runs on the previous generation of the same family (Sonnet 4.6, Opus 4.6) and is billed at that model's rate.

Quanty picks the cheapest model that does the job well. For AI columns that means Claude Haiku 4.5 for simple, bounded columns (classification, number, date, single and multi select, URL) and Claude Sonnet 5 for everything else (custom prompt, summary, entity extraction, key-value, JSON). You can override that per column from the column's model picker — including Opus 5 or Fable 5 when a column really needs them. In chat, the auto mode answers quick questions with Claude Sonnet 5 and hands complex work to Claude Opus 5. Lighter tasks go to the Amazon Nova models and gpt-oss; the internal helper steps — classifiers, naming, template matching, retrieval query rewriting and document overviews — run on Amazon Nova 2 Lite, Nova Lite or Nova Micro, which is why those models appear in the table even though you rarely pick them yourself. Because the default is not the flagship, an allowance goes several times further than it otherwise would. The GPT-5.6 models are no longer offered; a column that had one selected now runs on the equivalent Claude or Amazon Nova model automatically.

The 1.3 multiple covers every AI model we run for you — the language models above, plus the embeddings behind document search (Cohere Embed v4 on Amazon Bedrock) and the speech-to-text behind voice input (Amazon Transcribe), both of which now run on AWS as well. Two line items convert on their own terms: web search is billed at 2.5x the search provider's measured cost, and file storage costs 60 credits per GB-month.

Rates can go down as well. An earlier repricing took Claude Sonnet 5 from $3.00 / $15.00 to $2.60 / $13.00 per million tokens, and the move to Amazon Bedrock leaves every Claude rate where it was.

Our commitments

• Provider price cuts are passed through. When a model provider lowers its prices, our published rates follow automatically. The same credits then buy more work — we do not keep the difference.
• Changes are announced in advance. Any change to how fast AI work consumes credits is announced at least 30 days ahead, and never takes effect mid-billing-period.

Plans and the subscription gate

Plans (Basic, Pro and Enterprise) are chosen on the Choose your plan page. "Every plan includes a monthly allowance of AI credits." Plans also set the organization limits you see elsewhere: "Up to count team members", "Up to count workbooks" and "count GB of file storage".

An active subscription is required to use AI features at all. Without one, the app shows "Subscription required — Quanty requires an active subscription to use AI features" and anything that would spend credits is paused. Organization members are covered by the owner's subscription.

How the credit balance works

Your balance has two parts, spent in a fixed order:

1. Plan allowance — granted every billing period, resets on renewal. Used first.
2. Top-up balance — bought as one-time packs. Rolls over between periods and is only consumed after the monthly allowance is gone.

The AI credits card in Settings → Billing shows credits remaining, "used / limit used this period", the top-up balance and the renewal date.

Warnings appear automatically:

• "Running low — less than 10% of your monthly allowance left"
• "Credits used up — AI features are paused" (AI stays paused until you top up or the period renews)

Topping up

Top up credits offers one-time packs: "One-time packs. Top-up credits roll over and are used after your monthly allowance." Choosing Buy redirects to Stripe checkout; when payment completes, the credits land in your balance immediately and you receive a Top-up credited notification plus an email receipt.

What spends credits

Every AI feature debits the same balance:

• Chat messages and tool runs
• AI column extraction and reprocessing (Columns)
• Voice transcription (billed per audio minute)
• Phone calls answered by the AI Secretary: the voice provider's measured cost at the same 1.3 multiple, charged once per call
• Agents and agent task runs
• Generating and regenerating presentations and dashboards
• AI formula drafting (Formulas)
• Email classification in automations
• Live web searches run by internet-fed columns (Columns) — one search per processed row, shown as its own Web search line on the usage page
• File storage, metered per GB-month

Deterministic work is free: formula columns compute at zero credits, and cells excluded by a branch never run.

What typical operations cost

For scale: the 20,000 credits in Basic's monthly allowance are roughly 1,700 standard cells, or roughly 3,400 simple ones.

These are estimates. Real cost depends on the length of the documents, the prompt and the answer — more context means more model tokens, and so more Quanty credits. The actual figures for every run are in Settings → Usage and in the cell details (Reviews & audit trail).

The credit ledger

Recent credit activity lists every movement of your balance:

Organization budgets and member limits

Independent of the credit balance, owners can cap spending inside the organization (see Organizations & roles):

• Monthly AI budget — org-wide cap; when it is reached, AI actions are paused for the whole organization until the month rolls over or the owner raises the cap.
• Member AI spend limits — per-member monthly caps with month-to-date spend shown next to each person.

Notifications cover the important events: budget reached, member limit reached, seat limit reached, payment failed, subscription past due or canceled, credits running low, credits used up, and top-up credited.

Usage, invoices and company details

• Settings → Usage breaks down credit and cost usage over time. Owners and admins see every member; members see themselves.
• Invoices lists your Stripe invoices; Manage Subscription opens the Stripe billing portal for payment methods, plan changes and cancellation.
• Company details shows the billing name, address and tax IDs on file with Stripe.

Related pages

• Organizations & roles — budgets, limits and seats.