Transparent pricing
Pricing
Per-request gateway pricing. No per-seat tax. No markup on model tokens. You pay for routing — each model provider invoices you directly under your existing agreement.
Pricing tiers
Starter
$299
/mo
$249/mo billed annually
500,000 requests / month
- Up to 3 endpoint types
- 3 routing policies
- 7-day audit log retention
- OpenAI-compatible API
- Community support
- 14-day free trial
Most Popular
Team
$899
/mo
$749/mo billed annually
5,000,000 requests / month
- Unlimited endpoint types
- Unlimited routing policies
- 90-day audit log retention
- Tenant isolation
- Data class enforcement
- Slack support
- Private GPU support (up to 4 nodes)
Enterprise
Custom
Unlimited requests
- Everything in Team
- Unlimited private GPU nodes
- PrivateLink + VPC peering
- 1-year audit log retention
- SAML SSO
- Custom SLA
- Dedicated onboarding engineer
All prices USD. Token costs not included — invoiced directly by your model providers.
Starter and Team tiers are self-serve and include monthly request quotas; Enterprise is contact-only and includes private GPU options. We do not charge per-token markup — only per-request routing.
FAQ
Common questions
Each call to the /v1/route endpoint counts as one request, regardless of model or output length.
No. Kamiwaza charges only for routing. Your model provider invoices separately — we never mark up tokens.
Yes. Register any HTTP-compatible inference endpoint, including private GPU clusters via VPC peering or PrivateLink.
The Starter plan comes with a 14-day full-feature trial. No credit card required.