What it is
Reducto is an API for turning PDFs, scans, spreadsheets and slides into structured, citation-ready JSON for AI systems. Its endpoints are Parse, Extract (typed fields with citations and bounding boxes), Split, Classify and Edit (filling PDFs and Word files back in), plus a no-code Studio. It offers Python, Node and Go SDKs, a CLI and an MCP server, and handles more than 30 file types.
On the Standard pay-as-you-go plan, parsing with its r-1 model costs $10 per 1,000 pages, Extract and Split $20, Deep Extract $40 and Edit $60, with 20% off Parse and Extract in batch and $150 of free usage to start. Growth and Enterprise plans are custom-priced. Growth adds zero data retention, a HIPAA business associate agreement and EU or Australian data residency; Enterprise adds VPC or on-premises deployment, a custom SLA and SSO.
The company was founded in 2023 in San Francisco and came through Y Combinator's Winter 2024 batch.
Our review
Reducto sells accuracy on hard documents, and the evidence for it is stronger than usual for this market. On LongExtractBench, 225 documents averaging 358 pages, Deep Extract ranked first of seven systems with 99.6% precision and recall and no failures. Reducto commissioned that benchmark and wrote its grading method; micro1, which published it in July 2026, sourced the documents, verified the ground truth and reviewed the method. Its older RD-TableBench results are Reducto's own.
Parsing is priced like a mid-market tool. At $10 per 1,000 pages, standard parsing costs less than LlamaParse's Agentic tier ($12.50) and more than its Cost Effective tier ($3.75). Extraction is where the bill grows: Deep Extract at $40 per 1,000 pages and Edit at $60 add up quickly on long filings, and the 20% batch discount applies only to Parse and Extract.
The compliance features are split across tiers in a way buyers should map before signing. Reducto's docs describe SOC 2 Type II controls and a signed BAA, but the BAA, zero data retention and EU residency are listed under Growth, not the self-serve plan. SSO, role-based access and VPC or air-gapped on-premises deployment are Enterprise-only. An independent profile on IDP-Software notes that the hosted service uses frontier-model providers as subprocessors and that its multi-pass processing is slower than single-pass rivals.
Throughput is published, which is rarer than it should be: 200 guaranteed concurrent pages on Standard, 350 on Growth and 500 or more on Enterprise. For a bank or insurer extracting fields from 300-page filings, those limits and the on-premises option matter more than the per-page price. For high-volume retrieval over ordinary PDFs, LlamaParse or self-hosted Docling does the job for less.
Where it earns its place
A credit team at a mid-sized lender processing loan files
Pull covenants and financial figures from 300-page agreements with page-level evidence
Deep Extract returns typed fields with citations and bounding boxes so analysts check only flagged values. At $40 per 1,000 pages, a 300-page file costs about $12 at list price; Growth discounts are negotiated.
A platform engineer at a health-tech startup
Parse clinical PDFs into a retrieval pipeline under HIPAA
Prototype on Standard with the $150 credit, then move to Growth for the BAA and zero data retention, which deletes documents within 24 hours. Start-ups under $15 million in funding qualify for discounted Growth pricing.
A head of operations at an insurer with an on-premises mandate
Run document extraction without sending claims data to a vendor cloud
Enterprise offers full VPC or air-gapped on-premises deployment with SSO and a custom SLA. Pricing is custom and not published.
Before you commit
- Setup
- Create an API key, install the Python, Node or Go SDK and parse a first file; Studio offers a browser route without code.
- Integrations
- REST API, three SDKs, a CLI and an MCP server; more than 30 file types including scans, spreadsheets and slides.
- Data and privacy
- SOC 2 Type II and AES-256 encryption at rest; zero data retention (deletion within 24 hours), BAA and EU or AU residency on Growth; VPC and on-premises on Enterprise.
- Cost at scale
- One million pages a month costs $10,000 to parse at list price, or $8,000 in batch; Extract on the same volume costs $20,000, or $16,000 in batch, before Growth discounts.
- Limits
- Guaranteed concurrency of 200 pages on Standard, 350 on Growth and 500 or more on Enterprise; Standard includes up to five Studio seats.
What's good
- First place on a third-party-validated long-document extraction benchmark
- Citations and bounding boxes on extracted fields
- Published concurrency limits on every plan
- VPC and air-gapped on-premises deployment on Enterprise
- Python, Node and Go SDKs plus an MCP server
What's not
- Much of the accuracy evidence is vendor-commissioned
- Deep Extract and Edit are costly on long documents
- BAA and zero data retention need the custom Growth plan
- SSO only on Enterprise
- Slower multi-pass processing
Best for
Teams extracting fields from long financial, legal or medical documents where errors are expensive.
Alternatives
Reducto: common questions
- Is Reducto worth it?
- Reducto is the parser to pay for when an extraction error costs more than the page, though most of its accuracy evidence still traces back to Reducto. We rate it 4.3 out of 5.
- How much does Reducto cost?
- $150 free usage; Parse $10, Extract $20, Deep Extract $40, Edit $60 per 1,000 pages; batch 20% off Parse and Extract. Growth and Enterprise custom.
- What are the downsides of Reducto?
- Much of the accuracy evidence is vendor-commissioned; deep Extract and Edit are costly on long documents; bAA and zero data retention need the custom Growth plan; sSO only on Enterprise; slower multi-pass processing.
- Who is Reducto best for?
- Teams extracting fields from long financial, legal or medical documents where errors are expensive.
- How much does Reducto cost for a team or in production?
- One million pages a month costs $10,000 to parse at list price, or $8,000 in batch; Extract on the same volume costs $20,000, or $16,000 in batch, before Growth discounts.
- Is Reducto safe for company data?
- SOC 2 Type II and AES-256 encryption at rest; zero data retention (deletion within 24 hours), BAA and EU or AU residency on Growth; VPC and on-premises on Enterprise.
- What are the best Reducto alternatives?
- The closest alternatives we have reviewed are LlamaParse.
Answers checked Oct 23, 2026.
Make Reducto? Show this rating on your site
