Governance-first pricing.
Pick the scale that matches your team.
Every tier ships with the full DVARA AI governance platform — Policy-as-Code, immutable audit, PII redaction, cost attribution, and the DVARA MCP Proxy. Tiers differ in capacity, audit retention, and support level.
Trial
Free for 30 days. No credit card.
Solo
$59/mo flat. No per-request overage.
Starter
For small teams running production AI.
Growth
For teams running production AI at scale.
One isolated tenant per plan. Need more — prod/staging, per-client, or a private deployment? Consider self-hosting or managed hosting below.
Trial: one per company domain, not renewable — after 30 days continue on Solo or any paid tier.
Annual is billed once a year at 10× the monthly rate — 2 months free.
Every tier includes the full DVARA platform
Single SKU. Every governance capability gates on the license, not on the plan. Tiers differ only in monthly token capacity, audit retention, and support level — shown on the cards above.
- Policy-as-Code engine with dry-run, shadow, versioning + rollback
- Immutable HMAC-SHA256 signed + hash-chained audit
- PII + injection guardrails (regex always-on + Phileas opt-in, 17 filter types)
- Content filters, output-schema enforcement, grounding / hallucination detection
- OWASP LLM05 / LLM07 / LLM10 patterns; optional ML classifiers + HTTP plugins
- SOC 2 Type II / HIPAA / GDPR evidence packages on demand
- MCP Proxy: server registry, tool-call governance, PII scanning, credential centralisation
- Human approval gates enforced at execution; per-tenant tool / server rules
- Agent loop detection with auto-kill; multi-agent session tracking
- Full OpenTelemetry trace spanning LLM turns + MCP calls
- Real-time per-request cost attribution; budget caps per tenant / team / key
- Automatic model downgrade at soft limit; chargeback reports + cost forecasting
- Round-robin, weighted, capability-aware, latency-aware, cost-aware routing
- Circuit breakers, failover, retry; canary + A/B + shadow + multi-region
- Drop-in OpenAI-compatible API across all 14 providers
- Multi-tenant by design with row-level isolation; BYOK per tenant
- Streaming, structured outputs (Anthropic + Gemini included), vision
- Prometheus metrics + OpenTelemetry traces + structured JSON logs
Self-hosted
- Deploy DVARA in your own VPC or cloud — data never leaves your perimeter.
- Entitlement is a signed license key, validated locally — no phone-home, air-gap friendly.
- You run and scale the cluster; fully multi-tenant with per-tenant isolation.
- For strict data-residency, regulated, or air-gapped environments.
Managed hosting
- We run a dedicated DVARA cluster for you, in your region.
- No infrastructure to operate — upgrades, scaling, and availability are handled.
- Dedicated and isolated, separate from the shared self-service cluster.
- For enterprise scale and isolation without running it yourself.