Turn Paperwork into finished work.

The agentic paperwork platform to parse documents, transcribe speech, decide what happens next and act on it.

Send a PDF, a scan or a call recording; get structured values, typed decisions, filled forms and tamper-proof signatures.

From messy file to trusted result.

Most document APIs stop after parsing or extraction. CloudRaker carries documents and audio from messy input to trusted output under one surface.

Turn PDFs, scans, office files and call recordings into structured markdown, cited JSON and diarized transcripts with Parse, Split, Extract and Transcribe.

Explore Understand

Run your whole document pipeline on one API.

Run all your workflows across parsing, extraction, form filling and signatures, instead of wiring four vendors together. Hand this prompt to your coding agent and it maps where CloudRaker fits in your project.

from cloudraker.client import CloudRaker

client = CloudRaker(token="YOUR_API_KEY")

result = client.extract(
    file={"url": "https://example.com/invoice.pdf"},
    schema={
        "type": "object",
        "properties": {
            "vendor": {"type": ["string", "null"]},
            "total": {"type": ["number", "null"]},
        },
    },
    citations=True,
)
Response
{
  "object": "extract_run",
  "status": "processed",
  "output": {
    "value": {
      "vendor": "Northwind Supply",
      "total": 4850
    },
    "citations": {
      "vendor": [{
        "page": 0,
        "bbox": { "x": 0.08, "y": 0.11, "width": 0.24, "height": 0.02 },
        "text": "Northwind Supply",
        "confidence": 5
      }],
      "total": [{
        "page": 1,
        "bbox": { "x": 0.71, "y": 0.83, "width": 0.13, "height": 0.02 },
        "text": "$4,850.00",
        "confidence": 5
      }]
    }
  }
}
Read the API docs

Unmatched accuracy, unbeatable price.

Our document parsing model scores 80+ on ParseBench, outranking specialized document parsing platforms, frontier models, and legacy document intelligence systems.

80+
ParseBench v1 Score
$10
Per 1,000 pages parsed
$0.001
One credit across every primitive
Most attractive quadrant More accurate, more expensive Less expensive, less accurate Less accurate, more expensive 40 50 60 70 80 90 $1 $2 $5 $10 $20 $50 $100 $150 Price (USD per 1,000 pages, log scale) ParseBench overall score Frontier rakedoc-nano-2: 80.1 · $10 per 1,000 pages rakedoc-nano-2 rakedoc-nano-2 Enterprise: 80.1 · $5.2 per 1,000 pages rakedoc-nano-2 Enterprise Upstage Parse 2: 80.6 · $30 per 1,000 pages Upstage Parse 2 Opus 5.5: 78.8 · $61.3 per 1,000 pages Opus 5.5 GPT-5.6 Sol: 74.4 · $80.8 per 1,000 pages GPT-5.6 Sol Gemini 3 Flash: 74.1 · $19.8 per 1,000 pages Gemini 3 Flash Datalab Accurate: 72.9 · $10 per 1,000 pages Datalab Accurate Reducto (Agentic): 70.9 · $47.6 per 1,000 pages Reducto (Agentic) Mistral OCR 4 (Annotation): 68.2 · $5 per 1,000 pages Mistral OCR 4 (Annotation) Reducto r-1-mini: 66.1 · $1 per 1,000 pages Reducto r-1-mini Databricks: 70.1 · $3.5 per 1,000 pages Databricks LlamaParse Cost Effective: 77.7 · $3.75 per 1,000 pages LlamaParse Cost Effective LlamaParse Agentic: 82.8 · $12.5 per 1,000 pages LlamaParse Agentic LlamaParse Agentic Plus: 86.7 · $56.25 per 1,000 pages LlamaParse Agentic Plus Reducto: 67.8 · $23.8 per 1,000 pages Reducto Azure Document Intelligence: 56.8 · $10 per 1,000 pages Azure Document Intelligence Google Cloud Document AI: 48.7 · $10 per 1,000 pages Google Cloud Document AI AWS Textract: 48.3 · $15 per 1,000 pages AWS Textract Cohere Parse v5 (est.): 50.1 · $1.5 per 1,000 pages Cohere Parse v5 (est.)

ParseBench public leaderboard, including our submitted run. Published 9 November 2026.

See how we measure

Transcribe at 270x realtime.

rakeaudio-asr transcribes a 10-minute call in about 2 seconds, starting at $0.30/hour.

realtime
270 x
with speaker labels
150 x
Speed vs. realtime
  1. Deepgram Nova-3 300x
  2. rakeaudio-asr 270x
  3. Gemini 87x
  4. gpt-transcribe 42x
  5. 4o-mini 41x
  6. 4o 37x
  7. whisper-1 29x

One 10-minute two-speaker file, compared with the Artificial Analysis leaderboard. Deepgram is faster. Gemini 3.5 Transcribe is more accurate, at 2.6% WER against about 5% in our internal English test.

Connect your agent in one command.

Claude, Cursor, Codex and ChatGPT connect over MCP with OAuth. The CLI and API use the same key.

  1. Works in Claude Code, Codex and Cursor
    Preview the prompt
    Help me integrate CloudRaker, the Paperwork platform, into this project.
    Read https://docs.cloudraker.com/llms.txt. Append .md to any docs page for Markdown.
    Install the Paperwork skills for the coding agent you are running in:
    npx skills add cloudraker/cloudraker-paperwork-skills --agent <client> --skill '*' --yes
    Choose claude-code, codex, or cursor for <client>. The MCP server at mcp.cloudraker.com serves the whole API.
    
    1. Read this codebase and the process it supports. Find where documents, scans, call recordings, forms or signatures enter or leave it.
    2. For each place, name the primitive that fits (parse, split, classify, extract, transcribe, decide, fill, sign) and the surface: the Python SDK, the TypeScript SDK, the paperwork CLI, or MCP. Say which file or step changes and what it replaces.
    3. Flag manual steps that one call could replace, even where no code exists yet.
    4. Return a ranked list with location, primitive, rationale and effort. Ask me where to start before you write code.
    5. API keys are self-serve at https://signup.cloudraker.com, with $15 of free usage a month. Never print keys, and keep them out of the repository and chat.
    
    Context: home

Usage-based pricing for any volume.

Every account gets $15 of free usage a month, no card required.

  • Parse

    $0.001/page

    Markdown. $0.01/page for OCR with rakedoc-nano-2.

  • Transcribe

    $0.30/hr

    $1.20/hr with speaker labels.

  • Extract

    $0.015/unit

    Smart. $0.03/unit for Advanced.

  • Decide

    $0.04per 1M input tokens

    125M free test tokens/mo.

  • Sign

    $0/envelope

    3 open envelopes on Free.

Built for regulated paperwork.

SOC 2 Type 2, HIPAA and configurable data residency.

  • SOC 2 Type 2 and HIPAA. Certified controls for regulated data. The SOC 2 report comes with Pro and Enterprise.

  • Data residency. Pick the region on Enterprise, or run in your VPC.

  • Open weights. rakedoc-nano-2 is on Hugging Face under AGPL-3.0.

  • SSO and SAML. Enterprise signs in through your identity provider.

Read security and compliance

Your first workflow is one prompt away.

$15 of free usage a month covers 7,500 rule-based parses or 1,000 Smart extractions, with no card required.