Pricing

Your router learns,
audits, and optimizes
itself.

Provara is an intelligent LLM gateway that catches upstream model regressions automatically, A/B tests its own ambiguous routing decisions, and migrates to cheaper models when they reach parity. Bring your own API keys — we charge for the intelligence, not the tokens.

Start Free Compare plans ↓

No credit card to startBYOK — pay providers directlyCancel anytimeOpen-source self-host available

Plans

Start free, upgrade when you're ready. Every tier includes the core gateway.

Free

Try Provara Cloud with no credit card.

$0forever

10,000 requests/mo
Full dashboard + analytics
Adaptive routing
A/B testing
LLM-as-judge sampling
1 seat
No auto-A/B generation
No silent-regression detection
No auto cost migrations

Start Free

Pro

Individual developers and small side projects.

$29/mo

100,000 requests/mo
Auto-A/B generation
Silent-regression detection
Auto cost migrations
3 seats
Email support

Start Pro

Team

Growing teams shipping AI features.

$149/mo

500,000 requests/mo
Everything in Pro
10 seats
Priority support
Team invites + SSO-ready

Start Team

Enterprise

Volume, dedicated support, custom terms.

Custom

Custom request volume
Unlimited seats
SAML SSO
Dedicated SLA
Priority support channel
Custom invoicing

Contact Sales

Compare every feature

The full grid — what's in each tier, what isn't.

Feature	Free	Pro	Team	Enterprise
Core Gateway
OpenAI-compatible API
Multi-provider routing (OpenAI, Anthropic, Google, xAI, Mistral, Z.ai, Ollama, custom)
Classifier + adaptive routing
A/B testing (manual)
LLM-as-judge quality scoring
Semantic + exact cache
Guardrails (PII detection, content policies, regex)
Request logs + replay
Monthly request quota	10k	100k	500k	Custom
Overage billing	Hard cutoff	$0.50 / 1k	$0.50 / 1k	Custom
Intelligence
Auto-A/B generation — spawns experiments on tied cells automatically
Silent-regression detection with closed-loop rerouting
Auto cost migrations with projected savings
Background scheduler observability
Team & Admin
Seats	1	3	10	Unlimited
Team invites Coming soon
Role-based access (owner / member)
API token scoping + rate limits
Audit logs
SAML SSO
Support & SLA
Community support (GitHub Discussions)
Email support
Priority response
Dedicated support channel
Custom SLA
Invoice billing

Frequently asked

Can I self-host instead?

Yes. The core gateway is open-source and self-hostable — you get the full routing engine, judge, A/B testing, adaptive matrix, guardrails, and analytics. Intelligence features (auto-A/B, silent-regression detection, cost migrations) are Cloud-only. A commercial self-host license with support is available for enterprise teams that need isolation.

What happens if I go over my monthly request quota?

On Pro and Team, overage requests keep flowing — you're billed $0.50 per 1,000 additional requests. On Free, requests stop at the quota and you'll need to upgrade or wait for the next billing period. Overage is billed on the same invoice as your subscription; no separate surprise charge.

Do you mark up my provider costs?

No. You bring your own API keys for OpenAI, Anthropic, Google, etc., and pay those providers directly. Provara's pricing is for the routing, analytics, and Intelligence layer on top — not a margin on tokens.

Can I change plans anytime?

Yes. Upgrade or downgrade from the dashboard; changes prorate automatically. Annual plans get two months free versus monthly. Cancellations stop renewal at the end of your current period — no refunds for partial months, but you keep access until the period ends.

Is Provara SOC 2 certified?

Not yet. We're a young product. Security posture today: AES-256-GCM for API keys at rest, OAuth-only sign-in (Google and GitHub, both with verified email), tenant-isolated data, encrypted TLS everywhere. SOC 2 Type II is on the roadmap for 2026 H2 as Enterprise customers begin to require it.

How do I talk to a human about Enterprise pricing?

Email [email protected] — include your expected request volume, team size, and any compliance or SLA requirements. We'll get back within one business day with a tailored quote.

How is my routing data isolated on Team and Enterprise?

Team and Enterprise tenants default to an isolated routing matrix — your router reads only your tenant's quality scores, and your ratings update only your tenant's matrix. Two opt-in toggles let you consume or contribute to the shared pool. Enterprise customers get contractual commitments on isolation, audit logs, and deletion — see the Enterprise Data Handling Addendum at /enterprise-data-handling for the full text.

Ready to route smarter?