Start extracting
VERALENS

Turn anythinginto clean datain seconds

Turn messy data into structured data you can ship.
Built for production, not one-off chat prompts.
Catches mistakes before they reach your database.

Start extracting

Built for production.
Not just demos.

Veralens turns frontier models into reliable infrastructure for document extraction at scale.

Hallucination-minimized by design

Every field is verified against OCR and layout before it reaches your JSON.

Trusted output, every time

ISO-native outputs

Dates, currencies, and country codes are normalized to ISO 8601, ISO 4217, and ISO 3166 automatically.

Clean data without post-processing

Jobs that always finish

Extractions are queued, tracked, and retried intelligently until they succeed or fail, no silent drops.

Built for production pipelines

Async-first, serverless-ready

Long-running docs, retries with backoff, webhooks, and usage metering, so your systems stay fast and resilient.

Designed for scale

No training on your documents

Your files are processed for extraction only. We never fine-tune on customer uploads or require template training.

Your data stays yours

Schema or prompt, your call

Provide a JSON Schema for strict typed output, or use natural-language prompts for Markdown, JSON, or plaintext.

One API. Full flexibility

Only pay for what you use

Top up prepaid funds anytime.
No subscriptions and no expiring credits.

Recommended

vera-1.0-high

Highest accuracy for complex layouts, dense scans, and strict JSON schemas.

$1.25/ 1M input tokens
$4.50/ 1M output tokens
  • Best for invoices, receipts & forms
  • Schema-guided JSON extraction
  • Recommended default engine
Top up & start

vera-1.0-low

Lower cost for simpler documents and high-volume batch runs.

$0.75/ 1M input tokens
$3.75/ 1M output tokens
  • Ideal for batch image extraction
  • Same API, lower per-token cost
  • Switch engines per job
Top up & start

Usage is deducted per extraction based on input and output tokens.

Frequently asked questions

Everything you need to know about extraction, jobs, and billing.

General models excel at open-ended text. Production extraction needs schema control, OCR cross-checks, ISO normalization, and durable async jobs. Veralens wraps that into one API so you get structured output without building the pipeline yourself.
The API accepts image sources (PNG, JPEG, WebP, GIF, HEIC/HEIF) and plain-text sources. Outputs are JSON, Markdown, or plaintext. The Playground currently supports image uploads; PDF and other document types are coming soon.
Top up prepaid funds via Stripe (from $5). Pick vera-1.0-high or vera-1.0-low per job. Usage is deducted from your balance based on input and output tokens per extraction. Funds never expire and there are no subscriptions.
Jobs move through pending, then completed or failed — every job ends in a clear state with timestamps and duration. Transient inference errors retry automatically across upstream models. Poll GET /v1/jobs/:jobId/get or set webhook_url on create (API key auth only).
Job payloads and API keys are encrypted at rest; all traffic uses HTTPS. Your documents are processed for extraction only — we never train on customer uploads. Redacted or obscured content in source documents is preserved in output instead of guessed.
Yes. Use the REST API with async jobs and optional webhooks from any stack. Sign in, create an API key, try the Playground, then call POST /v1/jobs/create from your app.