{
  "listings": [
    {
      "slug": "docling",
      "name": "Docling",
      "company": "Docling Project (LF AI \u0026 Data Foundation)",
      "grade": "B",
      "fit": "yes",
      "why": "MIT-licensed library that runs locally on macOS, Linux and Windows, x86_64 and arm64, with a documented CPU-only torch install; there is no hosted version to differ from. acc-18's Windows install failure predates the current install page.",
      "source": "https://docling-project.github.io/docling/getting_started/installation/",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/docling"
    },
    {
      "slug": "xiaomi-ocr-0",
      "name": "Xiaomi-OCR-0",
      "company": "Xiaomi Corporation",
      "grade": "C",
      "fit": "yes",
      "why": "Apache 2.0 weights with a documented install; the example code falls back to CPU in float32, and community GGUF builds exist. There is no hosted version to differ from.",
      "source": "https://huggingface.co/xiaomi-research/Xiaomi-OCR-0/raw/main/README.md",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/xiaomi-ocr-0"
    },
    {
      "slug": "azure-document-intelligence",
      "name": "Azure Document Intelligence",
      "company": "Microsoft Corporation",
      "grade": "B",
      "fit": "caveat",
      "why": "Read and Layout v4.0 containers run on 8 CPU cores with 10 to 24 GB, but they must send billing data to an Azure endpoint, and hosted inputs are kept 24 hours with a delete call.",
      "source": "https://learn.microsoft.com/en-us/azure/ai-services/document-intelligence/containers/install-run",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/azure-document-intelligence"
    },
    {
      "slug": "datalab",
      "name": "Datalab",
      "company": "Endless Labs, Inc.",
      "grade": "B",
      "fit": "caveat",
      "why": "Marker, Surya and Chandra weights are free under $2M to $5M revenue thresholds, and acc-14 rated Marker best to self-host, but the supported on-prem container needs an Enterprise contract.",
      "source": "https://documentation.datalab.to/docs/on-prem/overview.md",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/datalab"
    },
    {
      "slug": "extend",
      "name": "Extend API + MCP",
      "company": "CrowdView Inc. dba Extend",
      "grade": "B",
      "fit": "caveat",
      "why": "BYOC deploys into the customer's own cloud VPC and zero retention is available on request, with no training by inference providers, but there is no laptop or single-GPU install with stated hardware.",
      "source": "https://docs.extend.ai/security/deployment-options.md",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/extend"
    },
    {
      "slug": "invofox",
      "name": "Invofox",
      "company": "Invofox, Inc. (Delaware), with a registered branch in Spain",
      "grade": "D",
      "fit": "caveat",
      "why": "The vendor briefing claims on-premises and VPC deployment and an opt-in zero-retention mode, but no install docs or hardware are published, and corrections feed model training.",
      "source": "https://developers.invofox.com/documentation/integrate-with-an-ai-agent.md",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/invofox"
    },
    {
      "slug": "llamaparse",
      "name": "LlamaParse API + MCP",
      "company": "LlamaIndex, Inc.",
      "grade": "B",
      "fit": "caveat",
      "why": "Self-hosted LlamaCloud runs on Kubernetes through a Helm chart on Enterprise plans only, and still calls OpenAI, Anthropic or Google models, so documents leave the cluster unless enterprise cloud models are set up.",
      "source": "https://developers.llamaindex.ai/llamaparse/self_hosting/index.md",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/llamaparse"
    },
    {
      "slug": "nanonets",
      "name": "Nanonets API + MCP",
      "company": "Nano Net Technologies Inc.",
      "grade": "E",
      "fit": "caveat",
      "why": "The open-source docstrange library runs locally, but acc-41 found local output much worse than the hosted version.",
      "source": "https://github.com/NanoNets/docstrange",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/nanonets"
    },
    {
      "slug": "nemotron-3-5-parse",
      "name": "Nemotron 3.5 Parse",
      "company": "NVIDIA Corporation",
      "grade": "B",
      "fit": "caveat",
      "why": "Open weights under OpenMDW-1.1 with documented Transformers and vLLM serving, but the model is built for NVIDIA GPUs with no CPU path documented.",
      "source": "https://huggingface.co/nvidia/Nemotron-3.5-Parse/raw/main/README.md",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/nemotron-3-5-parse"
    },
    {
      "slug": "reducto",
      "name": "Reducto API + MCP",
      "company": "Reducto, Inc.",
      "grade": "B",
      "fit": "caveat",
      "why": "Dedicated VPC and on-prem deployments are documented with their own changelog and security model, but they are enterprise contracts, and zero data retention is Growth and Enterprise only.",
      "source": "https://docs.reducto.ai/onprem/enterprise_deployment_options.md",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/reducto"
    },
    {
      "slug": "unstructured",
      "name": "Unstructured API + MCP",
      "company": "Unstructured Technologies, Inc.",
      "grade": "E",
      "fit": "caveat",
      "why": "The open-source library installs with pip or runs from multi-platform Docker images for x86_64 and Apple silicon, but CPU and GPU requirements and any differences from Transform's output are not documented in the README.",
      "source": "https://raw.githubusercontent.com/Unstructured-IO/unstructured/main/README.md",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/unstructured"
    },
    {
      "slug": "veryfi",
      "name": "Veryfi API + MCP",
      "company": "Veryfi, Inc.",
      "grade": "C",
      "fit": "unchecked",
      "why": "No self-hosted option appears in the SDK README, and the retention and training pages were not re-read this pass.",
      "confidence": "low",
      "url": "https://www.anchorterminal.com/tools/veryfi"
    },
    {
      "slug": "adobe-pdf-extract",
      "name": "Adobe PDF Services / PDF Extract API",
      "company": "Adobe Inc.",
      "grade": "C",
      "fit": "no",
      "why": "Hosted only. Assets are kept 24 hours with a delete endpoint, but nothing runs on the user's machine.",
      "source": "https://raw.githubusercontent.com/AdobeDocs/pdfservices-api-documentation/main/src/pages/overview/security.md",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/adobe-pdf-extract"
    },
    {
      "slug": "amazon-textract",
      "name": "Amazon Textract",
      "company": "Amazon Web Services, Inc.",
      "grade": "BB",
      "fit": "no",
      "why": "Hosted AWS service only; no self-hosted path.",
      "source": "https://docs.aws.amazon.com/textract/latest/dg/full.md",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/amazon-textract"
    },
    {
      "slug": "cloudflare-clairic",
      "name": "Clairic",
      "company": "Cloudflare, Inc.",
      "grade": "BB",
      "fit": "no",
      "why": "Hosted on Workers AI only; open weights are promised with no date. The model page does state zero data retention.",
      "source": "https://developers.cloudflare.com/workers-ai/models/clairic/index.md",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/cloudflare-clairic"
    },
    {
      "slug": "google-cloud-document-ai",
      "name": "Google Cloud Document AI",
      "company": "Google LLC",
      "grade": "BB",
      "fit": "no",
      "why": "Hosted Google Cloud service only; no self-hosted processor.",
      "source": "https://docs.cloud.google.com/document-ai/docs/processors-list",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/google-cloud-document-ai"
    },
    {
      "slug": "landingai-agentic-document-extraction",
      "name": "LandingAI Agentic Document Extraction",
      "company": "LandingAI, Inc.",
      "grade": "BB",
      "fit": "no",
      "why": "Hosted in the US or EU only; the docs index lists no self-hosted or on-premises install. Zero Data Retention is a Team and Enterprise setting.",
      "source": "https://docs.landing.ai/llms.txt",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/landingai-agentic-document-extraction"
    },
    {
      "slug": "mindee",
      "name": "Mindee API",
      "company": "Mindee SAS (Mindee Inc. outside the Eurozone)",
      "grade": "C",
      "fit": "no",
      "why": "Hosted only. Extracted data is kept 12 hours by default (24 at most), the original file is never stored to disk, and processing can be pinned to the EU.",
      "source": "https://docs.mindee.com/models/data-processing-policies.md",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/mindee"
    },
    {
      "slug": "mistral-ocr",
      "name": "Mistral OCR API",
      "company": "Mistral AI (RCS Paris 952 418 325)",
      "grade": "B",
      "fit": "no",
      "why": "Hosted API only; no self-hosted OCR model in the docs read. Zero data retention is available to eligible organisations.",
      "source": "https://docs.mistral.ai/llms.txt",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/mistral-ocr"
    },
    {
      "slug": "opendocrouter",
      "name": "OpenDocRouter",
      "company": "LlamaIndex, Inc.",
      "grade": "B",
      "fit": "no",
      "why": "Hosted router only. Without cache nothing is stored, and cached results last 24 hours with a DELETE, but pages go to third-party model providers.",
      "source": "https://www.opendocrouter.ai/docs.md",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/opendocrouter"
    }
  ],
  "marks": "yes: the docs show it does the job as the success test says. caveat: it does the job with a limit an agent should know. no: it doesn't do this job. unchecked: we couldn't read it.",
  "meta": {
    "attribution": "Anchor Terminal (https://www.anchorterminal.com)",
    "attributionRequired": true,
    "credit": "CC BY 4.0, credit Anchor Terminal and link the page or API address you took it from (https://www.anchorterminal.com/terms/#data)",
    "docs": "https://www.anchorterminal.com/docs/",
    "generatedAt": "2026-11-16",
    "license": "CC-BY-4.0",
    "method": "https://www.anchorterminal.com/benchmark/",
    "methodology": "0.8",
    "openapi": "https://www.anchorterminal.com/openapi.json",
    "preview": false,
    "run": "2026-10-01",
    "runLabel": "October 2026 research run",
    "terms": "https://www.anchorterminal.com/terms/#data"
  },
  "useCase": {
    "accounts": 5,
    "after": [
      "vector-search.local-embedded",
      "document-extraction.rag-ingest"
    ],
    "area": "web-data",
    "asks": [
      "parse these locally, they can't leave the machine",
      "which parser can I self-host",
      "run OCR on my own GPU"
    ],
    "before": [
      "file-storage.agent-upload"
    ],
    "blockers": [
      "needs a CUDA GPU the user lacks",
      "local model weaker than the hosted one with the same name",
      "parsed output kept by the vendor with no delete"
    ],
    "capabilities": [
      "docs.parse",
      "docs.ocr",
      "model.open-weights"
    ],
    "category": "document-extraction",
    "categoryName": "Document parsing \u0026 extraction",
    "checked": "2026-11-15",
    "companies": {
      "fit": 2,
      "fitWithLimit": 9,
      "tagged": 113
    },
    "competition": "few",
    "fits": [
      "docling",
      "xiaomi-ocr-0"
    ],
    "fitsWithLimit": [
      "azure-document-intelligence",
      "datalab",
      "extend",
      "invofox",
      "llamaparse",
      "nanonets",
      "nemotron-3-5-parse",
      "reducto",
      "unstructured"
    ],
    "id": "document-extraction.private-local-parse",
    "listings": {
      "caveat": 9,
      "no": 8,
      "notRead": 6,
      "tagged": 187,
      "unchecked": 1,
      "yes": 2
    },
    "name": "Parse documents without sending them out",
    "status": "draft",
    "success": "The tool runs on the stated hardware with a documented install, gives the same output as its hosted version where one exists, and the docs state retention, deletion and whether inputs train models.",
    "url": "https://www.anchorterminal.com/api/v1/use-cases/document-extraction.private-local-parse.json",
    "weight": 8,
    "weightBasis": "5 of 46 accounts (Docling, Marker, docstrange, LlamaParse retention); the Docling and Nemotron listings are built for it"
  }
}
