Simple, transparent pricing
Start free. Pay only when you need more. No surprises.
Free
For side projects and evaluation
- Up to 10,000 events/month
- 7-day data retention
- 1 user
- Community support
- Basic dashboard
- Privacy-first architecture
Pro
For teams tracking AI costs
- Up to 1M events/month
- 1-year data retention
- 10 users
- Email support
- Entity attribution (customer/team/project)
- Anomaly detection alerts
- CSV export
- API access
Enterprise
For regulated industries
- Unlimited events
- Custom retention
- Unlimited users
- Dedicated support
- SSO (SAML/OIDC)
- HIPAA BAA available
- SOC2 Type II report
- Custom integrations
- SLA guarantee
Frequently Asked Questions
What data do you collect?
Only metadata: token counts, model used, latency, cost, and your entity tags. We never see or store your prompts, completions, or any actual content. Your data goes directly to OpenAI/Anthropic.
Is InfraPrism HIPAA compliant?
Yes. Because we never see PHI (prompts/completions), we're compliant by architecture, not just policy. Enterprise customers can get a BAA if needed, but technically we never handle covered data.
What's an event?
An event is one tracked LLM API call. Each call to OpenAI or Anthropic counts as one event. Streaming responses count as a single event, not per-chunk.
What happens if I exceed my limit?
We'll notify you at 80% and 100% of your limit. Events beyond your limit won't be tracked, but your LLM calls continue working normally. We never block your application.
How is this different from Helicone/LiteLLM?
Those tools proxy your traffic - all your prompts flow through their servers. InfraPrism is SDK-only: we extract metadata client-side, so your actual data never touches us. This makes us suitable for regulated industries where proxy tools aren't an option.
Do you offer discounts?
Yes, we offer 20% off for annual billing, and 50% off for startups under $1M ARR and non-profits. Contact us to apply.
Ready to understand your AI costs?
Start tracking in minutes. No credit card required.