# Interfaze vs Reducto URL: https://interfaze.ai/compare/reducto [ Interfaze ![logo](https://interfaze.ai/interfaze_icon.svg) ](https://interfaze.ai/)[ we're hiring ](https://interfaze.ai/careers) models[ pricing ](https://interfaze.ai/pricing)[ help ](https://interfaze.ai/help)[ docs ](https://interfaze.ai/docs)[ blog ](https://interfaze.ai/blog)[ sign in ](https://interfaze.ai/dashboard/playground) Backed by![Y Combinator](https://interfaze.ai/yc_logo.svg)Combinator The whole OCR pipeline in one API call. One API call does OCR, classification and translation, and returns structured JSON. Every response also includes the OCR underneath, with bounding boxes and a confidence score for each word. [Build now ->](https://interfaze.ai/dashboard/playground)Setup with agents Trusted by builders at ![NVIDIA](https://interfaze.ai/companies/nvidia.webp) ![IBM](https://interfaze.ai/companies/ibm.webp) ![Vercel](https://interfaze.ai/companies/vercel.webp) ![Supabase](https://interfaze.ai/companies/supabase.webp) ![Notion](https://interfaze.ai/companies/notion.webp) ![Zapier](https://interfaze.ai/companies/zapier.webp) ![Robinhood](https://interfaze.ai/companies/robinhood.webp) ![Airwallex](https://interfaze.ai/companies/airwallex.webp) ![Langflow](https://interfaze.ai/companies/langflow.webp) ![Smithery](https://interfaze.ai/companies/smithery.webp) ![Rho](https://interfaze.ai/companies/rho.webp) ![Pocket](https://interfaze.ai/companies/pocket.webp) ![101 GenAI](https://interfaze.ai/companies/101genai.webp) ![Aegis](https://interfaze.ai/companies/aegis-logo-text.webp) ![RMH](https://interfaze.ai/companies/rmh-logo.webp) ![INTH](https://interfaze.ai/companies/inth.webp) ![Andco](https://interfaze.ai/companies/andco.webp) ## Why teams choose Interfaze over Reducto Bounding boxes and confidence scores you can check, the whole pipeline in one call, better reads on hard scans, and a smaller bill. [ INVOICE NO. INV-20418 DATE 2026-09-03 1,200.00 3,620.50 TOTAL $4,820.50 total\_due 0.98 01 ### Data you can verify Every response comes with bounding boxes and a confidence score for each word, so you can check any value against the page and flag the uncertain ones. ](https://interfaze.ai/docs/precontext) [ .pdf .jpg .mp3 URL ![](https://interfaze.ai/interfaze_icon.svg) 02 ### The whole pipeline in one call Parse, classify and translate in a single request. No extra APIs to wire up and no glue code to maintain. ](https://interfaze.ai/docs/vision/ocr) [ LINE ITEMS ITEM QTY AMOUNT Valve, 2in 12 1,104.00 Gasket kit 40 386.40 Labour, site 6h 720.00 TOTAL 2,210.40 Net 30. 1.5% monthly on overdue balances. Prices exclude VAT at the prevailing rate. 03 ### Better on hard-to-read scans On our hand-checked test set, interfaze-1 read more rotated, upside-down and faint text correctly than Reducto r-1, and its confidence scores caught every wrong answer. ](https://interfaze.ai/compare/reducto#benchmarks) [ Multi-step pipeline parse classify translate Interfaze one call, one bill 04 ### Cheaper per request One request covers the whole pipeline. On our test documents, Reducto cost about 30% more on average. ](https://interfaze.ai/pricing) [Try for free ->](https://interfaze.ai/dashboard/playground) ## Interfaze vs Reducto on the same document Same pages and same fields, sent to both APIs. These are pages where interfaze-1 got more fields right. Pick one to see where each API placed its fields, or the JSON it returned. Rotated table scan Dark cover scan Faint campaign form Boxes on the page Returned JSON correct value and box wrong value, broken text or bad box Interfaze · interfaze-1 Reducto · r-1 ![](https://interfaze.ai/interfaze_icon.svg)Interfazeinterfaze-1 structured JSON ![Typewritten table scan with handwritten corrections](https://r2public.jigsawstack.com/interfaze/compare/proof/pzjx0138_p1.jpg) ✓ usa\_total\_byproducts 25.6 0.99 ✓ germany\_light\_recon 22.3 0.99 ✓ venezuela\_bandcast 18.2 0.97 ✓ average\_europe\_total\_recon 22.7 0.99 ✓ malaysia2\_light\_recon 17.2 0.99 ✓ finland\_total\_recon 25.0 0.98 ✓ side\_number 597004904 0.99 ![](https://interfaze.ai/compare/reducto/reducto_icon.svg)Reductor-1 structured JSON ![Typewritten table scan with handwritten corrections](https://r2public.jigsawstack.com/interfaze/compare/proof/pzjx0138_p1.jpg) ✗ germany\_light\_recon 22.3 Right value, but the box is on the Switzerland row. 0.95 ✗ venezuela\_bandcast 18.2 Right value, but the box is on the 18.2 in the Total Recon column. 0.99 ✗ malaysia2\_light\_recon 17.2 Right value, but the box is on the 17.2 in the Total Recon column. 0.99 ✗ side\_number 597004905 · page says 597004904 0.86 ✓ usa\_total\_byproducts 25.6 0.99 ✓ average\_europe\_total\_recon 22.7 0.99 ✓ finland\_total\_recon 25.0 0.99 ## One API call instead of a pipeline [ OCR docs -> ](https://interfaze.ai/docs/vision/ocr) Both take a few seconds per page. With Interfaze you drop the extra APIs and the code that stitches them together. Global documents Claims intake Global documents integration steps Interfaze 1 call · 1 API waiting Reducto 0/5 steps · 0 APIs waiting With Interfaze With Reducto Scanned form 1 Parse Reducto 2 Classify Reducto 3 Translate LLM 4 Q&A LLM 5 Your code you ![](https://interfaze.ai/interfaze_icon.svg) Interfaze ocrclassifytranslateQ&A One request Structured output Interfaze 1 call · 1 API waiting Reducto 0/5 steps · 0 APIs waiting Scanned form 1 Parse Reducto 2 Classify Reducto 3 Translate LLM 4 Q&A LLM 5 Your code you ![](https://interfaze.ai/interfaze_icon.svg) Interfaze ocrclassifytranslateQ&A Structured output ## Interfaze vs Reducto accuracy benchmarks Where interfaze-1 beats Reducto r-1 on the same documents, checked by hand: confidence scores you can act on, boxes on the right cell, and hard-to-read text. Accuracy by content type higher is better Content ![](https://interfaze.ai/interfaze_icon.svg) Interfaze ![](https://interfaze.ai/compare/reducto/reducto_icon.svg) Reducto Errors flagged by confidence wrong answers scored below 0.9, so they reach review · 11 vs 10 errors 100.0% 20.0% Box on the right table cell field box on the cell the value came from · 8 table fields 87.5% 50.0% Rotated, upside-down & faint text upside-down pages, rotated margin numbers, faint stamps · 15 fields 73.3% 60.0% Fine print rotated numbers, stamps, footers, small type · 10 fields 80.0% 70.0% Body text names, titles, labels and running text · 24 fields 95.8% 91.7% [Method ->](https://interfaze.ai/leaderboards) ## Interfaze vs Reducto on olmOCR-bench [ Full breakdown -> ](https://interfaze.ai/leaderboards/olmocr) Scores from the public olmOCR benchmark. The overall ranking first, then the three tasks with the biggest gap: math in old scans, arXiv papers and multi-column layouts. ## Interfaze vs Reducto, feature by feature Checked against both companies' docs. ![](https://interfaze.ai/interfaze_icon.svg) Interfaze ![](https://interfaze.ai/compare/reducto/reducto_icon.svg) Reducto Structured JSON output Your JSON schema on any input, in the same call as the OCR (OpenAI json\_schema, Zod, Pydantic) Structured JSON output Separate Extract API, documents only, not OpenAI response\_format Multimodal input PDF, Word, scans and images, plus call recordings (timestamps, diarization up to 50 speakers) and video on interfaze-1 Multimodal input Documents and images only (PDF, Office, spreadsheets); no audio or video Bounding boxes with confidence scores Every OCR word and line with pixel bounds and a 0–1 confidence, included in every response Bounding boxes with confidence scores Per-field citations and word-level OCR with 0–1 confidence; OCR boxes cost extra on r-1 Open weights interfaze-1.1-lite runs on one 80 GB GPU Open weights No open weights; VPC and on-prem on Enterprise only Zero data retention Nothing kept after the request, on any plan, with one header Zero data retention Growth plan and above, and data is still kept for up to 24 hours Standard SDKs OpenAI Chat Completions compatible: OpenAI SDK, Vercel AI SDK, LangChain Standard SDKs Own SDKs only (Python, Node, Go); not OpenAI-compatible Classification and translation Same request as the OCR Classification and translation Classify is a second call; no translation endpoint in the API docs Streaming Streams the JSON object as it is generated Streaming Job-based: wait for the full result or poll Multi-document Q&A, analysis and comparison Send several documents in one request and ask questions across them, analyse or compare them Multi-document Q&A, analysis and comparison One document per Parse or Extract call; no question answering across documents in the API docs Video OCR Send a video to interfaze-1 and read the text and fields in it, like any document Video OCR No video input ## How to migrate from Reducto to Interfaze [ OpenAI SDK guide -> ](https://interfaze.ai/docs/integrations/openai-sdk) The same extraction in Interfaze and Reducto, side by side. Your existing schema works as is. Or let your coding agent make the switch. TypeScript TypeScript Python **OpenAI SDK · typescript** ```typescript import OpenAI from "openai"; const client = new OpenAI({ baseURL: "https://api.interfaze.ai/v1", apiKey: process.env.INTERFAZE_API_KEY, }); const res = await client.chat.completions.create({ model: "interfaze-1", response_format: { type: "json_schema", json_schema: { name: "invoice", schema: INVOICE_SCHEMA }, }, messages: [{ role: "user", content: [ { type: "text", text: "Extract the fields" }, { type: "file", file: { filename: "invoice.pdf", file_data: pdfUrl } }, ], }], }); const invoice = JSON.parse(res.choices[0].message.content); ``` **Vercel AI SDK · typescript** ```typescript import { generateText, jsonSchema, Output } from "ai"; import { interfaze } from "@interfaze-ai/ai-sdk"; const { output } = await generateText({ model: interfaze("interfaze-1"), output: Output.object({ schema: jsonSchema(INVOICE_SCHEMA), }), messages: [{ role: "user", content: [ { type: "text", text: "Extract the fields" }, { type: "file", data: pdfUrl, mediaType: "application/pdf" }, ], }], }); ``` **OpenAI SDK · python** ```python import json, os from openai import OpenAI client = OpenAI( base_url="https://api.interfaze.ai/v1", api_key=os.environ["INTERFAZE_API_KEY"], ) res = client.chat.completions.create( model="interfaze-1", response_format={ "type": "json_schema", "json_schema": {"name": "invoice", "schema": INVOICE_SCHEMA}, }, messages=[{ "role": "user", "content": [ {"type": "text", "text": "Extract the fields"}, {"type": "file", "file": {"filename": "invoice.pdf", "file_data": pdf_url}}, ], }], ) invoice = json.loads(res.choices[0].message.content) ``` **Interfaze SDK · python** ```python import json from interfaze import Interfaze, inputs, response_format interfaze = Interfaze() # reads INTERFAZE_API_KEY res = interfaze.chat.completions.create( messages=[{ "role": "user", "content": [ {"type": "text", "text": "Extract the fields"}, inputs.file(pdf_url, filename="invoice.pdf"), ], }], response_format=response_format(INVOICE_SCHEMA, "invoice"), ) invoice = json.loads(res.choices[0].message.content) ``` Migrate with agents ## Interfaze vs Reducto: common questions What people usually ask before switching from Reducto. Interfaze vs Reducto: what's the difference? Reducto is a document parser. Parsing, extraction and classification are separate endpoints, and audio or translation need another service. Interfaze is a general-purpose model: one call does OCR, classification and translation and returns structured JSON, for documents, images, audio and video. Will accuracy drop if I switch? Try your own documents on both first. The side-by-side and benchmarks above show where each API gets fields right and wrong on the same pages, checked by hand. Every Interfaze response also includes a confidence score for each word, so you can send uncertain values to review from day one. How much code do I need to change? A few lines per call. Point the OpenAI SDK or Vercel AI SDK at Interfaze, send the PDF as a file and pass the same JSON schema you use with Reducto Extract. You can delete the upload and job-polling code. Can I self-host? Yes. interfaze-1.1-lite has open weights and runs on a single 80 GB GPU (compute capability 8.9 or newer, like an H100). What about security and data retention? Turn on zero data retention for any request with one header, on every plan. SOC 2 and HIPAA agreements are available on request. Can Interfaze handle audio and video too? Yes. Speech-to-text with timestamps and speaker labels for up to 50 speakers, plus video input on interfaze-1. Reducto's docs list no audio or video formats. ## Send us your hardest PDF. If we can't beat your current provider, we'll give you $3,000 in credits. Start for free. No credit card required. [Build now ->](https://interfaze.ai/dashboard/playground)[Talk to a founder ->](https://interfaze.ai/sales)