Pricing
Simple pricing. No observability tax.
Bring the provider keys you already have — you pay for those tokens directly, no markup. Reach every other model through VernaOne’s own inference, metered in credits. No need to open an account with ten providers.
Free
Ship your first prompt as an endpoint.
- Router in the request path — 15+ models & tools
- Versioned prompt registry, promote & rollback
- Automatic fallback across providers
- Side-by-side compare (up to 3 models)
- Bring your own keys — routing always free
- 1 seat · community support
Pro
For engineers running LLM features in production.
- Everything in Free, plus:
- Active cost control — route to cheapest model that passes
- Cost & quality analytics per prompt version
- Lossless token compression (28–49% on data-heavy prompts)
- Eval checks + promote gates
- Unlimited compare · flows & agents
- Up to 3 seats · email support
Team
For teams shipping together.
- Everything in Pro, plus:
- Shared workspace & roles
- Up to 10 seats included
- Model-decommission alerts across all prompts
- Audit log & spend controls
- Priority support
Enterprise
For scale, security, and data residency.
- Everything in Team, plus:
- Self-host / private deployment option
- SSO / SAML & SCIM
- OpenTelemetry export (gen_ai.*)
- Security review & SLA
- Dedicated support & onboarding
How pricing works
Two ways to reach a model. You don’t need an account with every provider.
Bring the keys you already have — and reach everything else through VernaOne’s inference. One platform, every model, one bill.
Your keys → executions
Plug in the provider keys you have — OpenAI, Anthropic, whatever you’re on. You pay those providers directly and we never mark them up. Metered as executions against your plan.
Token markup: $0VernaOne inference → credits
Only have OpenAI, but want to try DeepSeek, Grok, or Claude? You don’t need to sign up for any of them. Run any model — and every AI-assist — on VernaOne’s inference, metered in credits.
Credits from $1/mo · top up anytime- Running or comparing a model you don’t have a key for — DeepSeek, Grok, Claude, media models…
- Prompt Optimize & Lab Assistant ≈ 3¢
- Explain-this-run ≈ 12¢ · Compose ≈ 36¢
- Proactive scouting for a cheaper model at equal quality across providers
Your own keys are never metered as credits — credits only cover models and assists that run on VernaOne’s inference.
Coming next: run your entire production workload on VernaOne inference — every model, no provider accounts at all.
Compare plans
What’s in each plan
| Capability | Free | Pro | Team | Enterprise |
|---|---|---|---|---|
| Executions / month (your keys) | 1,000 | 10,000 | 50,000 | Custom |
| Inference credits / month | $1 | $5 | $20 | Custom |
| Reach models without a provider account | ✓ | ✓ | ✓ | ✓ |
| Router in the request path | ✓ | ✓ | ✓ | ✓ |
| Versioned prompt registry | ✓ | ✓ | ✓ | ✓ |
| Automatic fallback | ✓ | ✓ | ✓ | ✓ |
| Bring your own keys (routing free) | ✓ | ✓ | ✓ | ✓ |
| Active cost control & routing | — | ✓ | ✓ | ✓ |
| Cost & quality analytics | Basic | ✓ | ✓ | ✓ |
| Token compression | — | ✓ | ✓ | ✓ |
| Eval checks + promote gates | — | ✓ | ✓ | ✓ |
| Seats included | 1 | 3 | 10 | Custom |
| Shared workspace & roles | — | — | ✓ | ✓ |
| Self-host / SSO / OTel export | — | — | — | ✓ |
Start free. Bring your own keys.
Route, version, and observe every LLM call from one place — no credit card, no per-trace tax.
Launch VernaOne →Pricing questions
How does billing work?
There are two meters, split by whose key runs the call. With your own provider keys, calls are counted as executions against your plan allowance and you pay the provider directly — no markup. To reach a model you don’t have a key for, or to use an AI-assist, the call runs on VernaOne’s inference and draws from your inference credits. Each plan includes a monthly execution allowance plus a credit allotment; top up credits anytime.
Do I need an account with every provider?
No — that’s the point. Bring the keys you already have (say, just OpenAI) and reach every other model — DeepSeek, Grok, Claude, media models — through VernaOne’s inference on credits. No signing up with ten providers, no ten invoices.
What counts as an execution?
One run of a prompt or endpoint on your own keys — a call to verna.run, a compare run, or a flow step that hits a model you’re subscribed to. Analytics, logging, versioning and traces are included and never metered separately.
What are inference credits, and what uses them?
Credits cover anything that runs on VernaOne’s own inference: running or comparing a model you don’t have a key for, and AI-assists like Optimize (~3¢), Explain (~12¢) and Compose (~36¢), plus proactive scouting for a cheaper model at equal quality. Credits are dollar-denominated (1 credit ≈ $1), so you always know what you’re spending.
Do you mark up my token costs?
On your own keys, never — you pay the provider directly and we don’t touch it. On VernaOne inference (models you don’t have a key for), you pay through credits at cost plus a small margin — the price of not having to run ten provider accounts yourself.
What happens if I run out?
We flag it before you hit a cap. Over the execution allowance, upgrade to the next plan; out of credits, top up or upgrade. Production calls on your own keys keep running regardless — only VernaOne-inference features pause until you top up.
Can I change plans or cancel anytime?
Yes — upgrade, downgrade, or cancel anytime from the dashboard. Bringing your own keys means you’re never locked in.