Simple pricing.
Powerful data.
Pay for what you use. One price covers labeled data, feature store API calls, model access, and LLM post-training engagements — no separate bills per service.
Get started with no commitment. One dataset, one model, limited API calls.
- 100 labeled samples/month
- Phishing URLs dataset only
- Daily refresh
- CSV download
- 500 Feature Store API calls/month
- Domain intelligence only
- Shared rate limit
- 1 ML model (URL Phishing Detector)
- 100 predictions/month
- Hosted API only
- Not included
For ML teams building their first cybersecurity model. Three datasets, five feature types, three models.
- 10K labeled samples/month
- 3 datasets (Phishing URLs, PE Malware, Browser Extensions)
- Daily refresh
- CSV and Parquet download
- Confidence scores included
- 50K Feature Store API calls/month
- All 5 feature types (WHOIS, PDNS, cert, content, domain)
- Batch extraction up to 1K domains
- Priority rate limit
- All 4 ML models
- 50K predictions/month
- Hosted API or export ONNX weights
- Auto-retrained on new data
- Available as a paid add-on engagement
- Frontier + small OSS model spot-check included
For production security teams. High-volume data, all models, custom labeling included.
- 50K labeled samples/month
- All 6 datasets
- Daily refresh cadence
- All formats (CSV, Parquet, JSON Lines)
- Custom labeling service (your data, private)
- Analyst reasoning notes per label
- 500K Feature Store API calls/month
- All 5 feature types
- Batch extraction up to 10K domains
- Dedicated rate limit
- Webhook notifications on data updates
- All 7 models (ML + DL)
- 500K predictions/month
- Hosted API, ONNX, or PyTorch export
- Fine-tuning on your labeled data
- Custom model retraining schedule
- One scoped post-training engagement per quarter included
- Priority scheduling
- Harness-building support included
All plans billed monthly. Cancel anytime. Need a custom volume deal? Contact us.
What's in each plan.
Labeled Data
Feature Store
Model Garden
Post-Training
What counts as a 'labeled sample'?
One labeled sample is one record from our dataset — a URL, a PE binary hash, a browser extension, etc. Downloading a dataset counts toward your monthly sample limit. If you re-download the same dataset, samples don't double-count.
What counts as a Feature Store API call?
One call per domain or URL submitted to any /v1/features/* endpoint. A batch request with 100 domains counts as 100 calls.
Are ML and DL predictions priced differently?
No. A prediction is a prediction, regardless of model type. DL models cost slightly more to host (reflected in the plan price), but there's no per-call surcharge.
My data is sensitive — will PreIntel see it?
Custom labeling is fully private. Your data is only accessed by the analyst assigned to the task, processed on isolated infrastructure, never shared, and never used to train our shared models.
Can I mix and match services?
Not yet — plans are bundled for simplicity. Enterprise customers can contact us for custom splits (e.g., more data, fewer API calls).