CapyParse Pricing 2026

One OCR API over nine hosted models — typed blocks, markdown, and schema extraction with citations

Pricing: Prepaid credits with per-page rates published per model and $10 of free credits at signup, no subscription tier. The cheapest engine, PaddleOCR-VL 1.6, is $1 per 1,000 pages; Azure DI Read is $1.73; Mistral OCR 3 is $2.30; Mistral OCR 4 is $4.60; Azure DI Layout is $11.50. Parse meters pages at the retail rate of whichever ocr_model the request names, and option surcharges are published live on GET /models/ocr. The /extract endpoint adds LLM token costs billed at the rate of the llm_model selected, on top of the page rate. Prices are in US dollars.

Document and email parser with four engines (AI, GPT, template, OCR) exporting to Sheets, CRMs, databases and webhooks.

Pricing: Free forever tier with 30 parsing credits per month. Paid tiers are billed annually: Starter from $24/mo ($290/year) for 100–1,000 credits, Growth from $124/mo ($1,490/year) for 5,000 credits, Business from $249/mo ($2,990/year) for 12,000 credits with volume options to 100k. All four parser engines, unlimited team members and priority support are included on every paid tier.

Template-free document data extraction to CSV, Excel or JSON with prepaid page credits valid for a year

Pricing: Prepaid page credits, no subscription and no auto-renewal. One credit processes one page, so a ten-page PDF costs ten credits. Packs start at $19 and scale through 100, 250, 500, 1,000, 2,500, 5,000, 10,000 and 25,000 pages; the published best-value pack is 1,000 pages for $99, which works out to $0.099 per page and is described as a 48% saving on the base rate, with the largest packs falling to roughly $0.051 per page. All credits are valid for twelve months from purchase with no monthly reset, and multiple packs can be bought and stacked. Every feature — all document types, batch processing and full API access — is unlocked at every tier rather than gated behind a higher plan, and there are no overage charges; the system alerts when the balance runs low. Signup grants 20 free page credits with no credit card.

Layout-aware document parsing that turns PDFs and images into RAG-ready structured data, pay-per-accurate-result credits from about $1 per 1,000 pages

Pricing: Credit-based and consumption-metered rather than tiered: the site's own words are pay only for what you use, and only for accurate results. A 7-day free trial grants $10 worth of credits (5,000 credits) to test against your own documents, and the first 30 users are offered a 20% lifetime discount on their plan. A pay-as-you-go pack starts at 6,000 one-time credits, with the parsed dataset retained for one month; metered billing is offered for enterprise-scale consumption. The vendor's published comparison table puts its own rate at approximately $1 per 1,000 pages, against $4–30 per 1,000 for Claude, $2–30 for unstructured.io and $1–20 for LlamaParse. A 30-day unconditional full refund is offered, processed within 24 hours of contacting support.

Native (non-OCR) PDF bank statement converter to Excel, CSV and QuickBooks .QBO, priced per transaction

Pricing: Pay-as-you-go transaction packs with no subscription and no monthly minimum; one transaction is one line on a statement, and a typical monthly statement holds 50–300. 200 free operations with no card. Starter is $1.99 for 1,000 transactions ($0.0020/txn); Basic $3.99 for 2,500 ($0.0016/txn); Standard $5.99 for 5,000 ($0.0012/txn); Business $9.99 for 10,000 ($0.0010/txn); Pro $19.99 for 25,000 ($0.0008/txn). Every pack includes direct .QBO import, 17 structured fields and the universal PDF-to-Excel parser. Prices in USD, with VAT or sales tax added at checkout where applicable.