01Model world
AI-routed OCR · live across every meridian
Planet 
CR
OCR reads. Planet decides what happens next.
- No credit card
- 1 free document · Advisor
- Asteroid from $0.003 / page
4 engines in orbit
AI route decide
JSON · MD · Text signal out
Upload
Drop anywhere
Route
AI selects best engine
Extract
Structured output
Deliver
Fast. Reliable. Global.
Score
Quality in real time
Every model is a world
Four engines, each tuned for a different document class. The router picks the fit — you get structured text out.
- 01
Asteroid Essential
Forms, invoices, and everyday docs
- 02
Quasar Enterprise
Long reports and research papers
- 03
Magnetar Professional
IDs, KYC packs, and handwriting
- 04
Pulsar Professional
Tables and complex layouts
Every Page Flies through the network
Hold to accelerate. Watch documents stream past engines, guardrails, and live routes.
Land structured data in your stack
The flow does not stop at a preview. Planet OCR ships the same engines through dashboard download, REST jobs, and API keys — with Guard watching the stream.
Structured payloads
Jobs return JSON, Markdown, or plain text — layout, tables, and extracted fields ready to drop into your pipeline.
REST API & keys
Upload a document, submit a job, poll until complete. Authenticate with an ocr_ key from the dashboard.
Guard in the stream
Refusals catch and reroute mid-flight so a weak page does not silently become a bad extract.
Engine routing
Asteroid, Pulsar, Magnetar, and Quasar stay in orbit. The advisor picks the fit; you can pin a model when you know.
Complete flow
From signup to structured data
Live product screens — Advisor through Upload, IDP, Ask Document, Reviews, Wallet, and Webhooks
Live workspace
Advisor
0:00 / 1:04
AdvisorChat your use case — volume, docs, API vs web.
Our Engine Fleet
Four models, built by us
Each engine in our constellation is purpose-built for different document challenges. The router picks the best fit — you get the result.
Asteroid
Multi-task powerhouse
Our foundational model trained for high-volume document processing. Handles tables, charts, formulas, and seals with precision across any scan quality.
Tables & chartsSeal recognitionFormula extractionBatch processing
0.003
$ per page
Quasar
Scientific & handwriting specialist
Our most advanced model for complex documents. Trained specifically for scientific papers, equations, multi-column layouts, and handwritten text recognition.
Handwriting OCRLaTeX equationsMulti-column layoutsScientific PDFs
0.008
$ per page
Pulsar
Speed-optimized parser
Built for velocity without sacrificing quality. Optimized for clean document parsing with state-of-the-art layout analysis and Markdown output generation.
Layout analysisMarkdown outputChart parsingFast inference
0.01
$ per page
Magnetar
Cloud-native intelligence
Designed for enterprise document workflows. Transforms complex scans into clean, structured Markdown with intelligent formatting preservation.
Clean MarkdownFormat preservationEnterprise scaleCloud-optimized
0.02
$ per page
All engines trained and optimized in-house. One API, one SDK — every model at your command.
Documents that survive the pull
Invoices, contracts, IDs, receipts — each page leaves as structured signal, already routed and scored.
- 01
Invoices
Line items, totals, and vendor fields
- 02
Contracts
Clauses extracted into structured text
- 03
IDs & passports
KYC fields without retyping
- 04
Filings & tables
Layouts kept, cells recovered
- 05
Receipts
Merchants, dates, and amounts
Images Become Information
Every document enters a living, intelligent network of specialized OCR engines. Planet OCR continuously benchmarks each model, understands the structure and complexity of your document, and automatically routes it to the engine best suited for the job.
If a model refuses the file, produces unreliable results, or fails during processing, Planet OCR instantly recovers by switching to a stronger alternative.