Skip to content
Pricing

Your needs pick the tier; your traffic sets the price

The platform fee is per seat, inference is per token consumed. Because the shape of your traffic drives the number, we quote it against a concrete usage profile rather than a list price.

Free

A single developer, first integration

To get to know the gateway on real traffic. A monthly trial credit, one region, basic guardrails.

  • Monthly trial credit
  • One region
  • Basic guardrails
  • Best-effort support

Pro

A small team already in production

Per-seat platform fee, Turkish home region, policy engine and combos switched on.

  • Policy engine + shadow mode
  • Combos and failover
  • Decision traces
  • 99.5% SLA
Most chosen

Team

Several teams, several products

Multi-region, budget hierarchy and the full guardrail pipeline. Prompt registry and canaries included.

  • Multi-region operation
  • Budget hierarchy
  • All guardrails
  • 99.9% SLA

Enterprise

Regulated sectors with audit obligations

Custom Rego modules, cross-region, custom retention and consulting on policy authoring.

  • Custom Rego modules
  • Cross-region
  • Custom retention
  • 99.95%+ SLA

Share your monthly request volume, average token length and compliance requirements; we come back with a firm quote within one business day.

Comparison table

Hover a row to see how the tiers differ

Comparison table
CapabilityFreeProTeamEnterprise
Access and models
One OpenAI-compatible endpoint
Model catalogCore poolWide poolFull catalogFull catalog + custom
Virtual API keys110UnlimitedUnlimited
Home regionOne regionTurkeySelectableCross-region
Attach your own vLLM cluster
Streaming and tool calls
Policy and compliance
YAML rule sets
Shadow mode and divergence report
Custom Rego modules
Guardrail pipelineBasicBasicFullFull + connectors
PII redaction obligation
Residency enforcement
Routing
Combos (failover chains)
Weighted splits
Decision graphs
Circuit breaker
Semantic cache
Prompt management
Prompt registry and versioning
Label-based publishing (Dev/Staging/Prod)
Canary traffic
Eval gate
Observability
Decision digest header
Full decision trace7 days30 days90 daysCustom
Cost breakdownBasicBy keyAll dimensionsAll dimensions
Conversation logs
OpenTelemetry export
Budget and administration
Per-key budgets
Budget hierarchy
Role-based access
SSO / SAML
Support
SLABest-effort99.5%99.9%99.95%+
Support channelCommunityEmailPriorityDedicated channel
Policy-authoring consultingLimited
Migration support
On-prem licensingOn request

About pricing

Because most of the total cost comes from the tokens you consume rather than the platform fee, and that depends on the shape of your traffic: your average prompt length, the share of requests that can be cached, the share that can drop to a cheap model. A list price quoted without seeing your usage profile misleads either you or us. Share the profile and we'll give you a firm number.

Inference is billed through Modelion: a margin is added on top of the provider cost. Models that require regional residency may carry an additional residency premium. Input and output tokens are priced separately.

Yes — moving up takes effect immediately. Moving down waits for the end of the period, and if you're using capabilities the new tier doesn't include, we tell you before the change.

The free tier is designed for getting to know the gateway on real traffic: a monthly trial credit, one region, basic guardrails. The policy engine and combos open up from Pro onwards.

On enterprise plans the inference margin can come down in exchange for committed use. We work the details through with sales.

Customers in Turkey are invoiced with VAT included and the e-invoice flow is supported. In other regions, billing runs through the relevant legal entity.