Pricing
Start with a bounded evaluation policy and promote to production when ready. Pricing is not token-based.
Bounded EVAL policy for benchmarking with receipts and hard limits.
- Hard caps (depth, steps, TTL)
- Hash-only by default
- Receipts + lineage for comparability
- Customer-hosted inference (no GPU hosting)
For teams running production workflows and vertical training.
- LIMITED_PROD policy profile
- Org quotas + audit logs
- Metrics dashboard
- Training compiler access
- Onboarding assistance
For organizations with specific requirements.
- Everything in Team
- Custom SLAs
- Dedicated support
- Security review
- Custom integrations
Common Questions
How do I evaluate LE Runtime?
Start with the EVAL policy profile via API. It enforces hard caps and returns deterministic receipts for side-by-side benchmarking.
Do you host inference?
No. LE is a control plane. Inference stays in your vLLM / sglang / HF environment. We provide policy enforcement and receipts.
Do you offer discounts for startups?
Yes. Early-stage companies building on LE Runtime may qualify for discounted pricing. Contact sales to discuss.
What is the Vertical Training Compiler?
A local compiler + SaaS control plane that turns vertical corpora into reusable spines and minimal deltas, reducing tokens and variance without sending raw data externally.