Pricing

Simple plans, one product

Seats cover the Prizmal Service. Routed calls are metered. Open and specialized models are passed through at cost.

One flat price: $50 per seat per month on every self-serve plan. The homepage calculator models that same $50 and still shows a 66% saving at the default profile.

Individual

For solo builders exploring managed inference

$50
per seat, per month
Routing fee: $0.75 per 1,000 API calls
1 seat
  • Prizmal managed inference
  • One key, OpenAI compatible
  • Live ROI dashboard
  • 99.9% availability target
  • Hard spend caps
  • SOC 2 Type II program underway
  • No lock in
  • Community support

Team

Popular

For small teams shipping together

$50
per seat, per month
Routing fee: $0.75 per 1,000 API calls
5 seats

Everything in Individual, plus:

  • Shared workspace and keys
  • Role-based access control
  • Team usage dashboards
  • Email support

Enterprise

For organizations that need control

Custom
volume seat pricing
10+ seats

Everything in Team, plus:

  • Managed inference
  • Routing control and policies
  • Governance, SSO, audit
  • Priority support and SLA

Local

Dedicated deployment in your region

Contact us
Custom
  • Dedicated tenant or on prem
  • Regional data residency
  • Custom model catalog
  • Named support engineer

Included in every plan

The full observability stack so your team can see what runs, why, and what it delivers

  • Tracing
  • Logs
  • Usage dashboard
  • Value dashboard

Serverless Inference

Optional add on

Available on every plan. Pay per request for on demand capacity with no infrastructure to manage

  • Per request billing, no minimums
  • Same routing and governance as your plan
  • Scales with your traffic, in and out

Questions we get

What does the routing fee cover?
Every routed API call passes through policy, tracing, and logging. The fee is $0.75 per 1,000 calls, billed on top of seats.
Do I have to change my code?
No. Prizmal is OpenAI compatible. Point your base URL at api.prizmal.ai and keep your SDK.
What availability do you commit to?
A 99.9% availability target, with hard spend caps and no lock in.
What does a seat cost?
A flat $50 per seat per month, on every self-serve plan. The homepage calculator models exactly that $50, and at the default profile of 50 seats and 66M tokens per seat per day it shows a 66% saving against a frontier baseline.
Is there more than one product to buy?
No. There is one product, the Prizmal Service: the decision layer, the runtime, and the Prizmal Controller for policy, identity, audit, and the kill switch.
Do you retain or train on my data?
No. Zero retention, never trained on. The details are on the Trust page.

See it on your own traffic

Bring your workload, keep your SDK, and see the savings on your own traffic