LLM gateway pricing: pay for the control plane, not the tokens

Inference runs on your own provider credential, at the provider’s list price. What you pay for is the control plane — failover, spend ceilings, guardrails and a ledger you can audit.

Two things unlock a request, and credits are not one of them

You need an API key from this dashboard, and your own credential for the provider that serves the model you picked — every model with a price runs on your key, not ours (where this deployment offers it, the free models also answer on ours until you add one, at $0, within a daily limit). Your provider bills you for inference directly, which is why a credit balance has nothing to pay for and is not sold today.

What you’re charged, and what you aren’t

Charged

Not charged

Solo — $0

0% BYOK fee, no credits to buy — your provider bills you for inference directly

Team · most adopted — $99

per month + 0% BYOK fee

Enterprise — Custom

annual, invoiced + 0% BYOK fee

All prices in USD

Compare the plans

FeatureSoloProTeamBusinessEnterprise
Price$0$29 a month per organisation$99 a month per organisation$399 a month per organisationCustom, annual
Free trialNot includedNot included14 days, no cardNot includedNot included
Markup on tokens0%0%0%0%0%
Unlimited API keysIncludedIncludedIncludedIncludedIncluded
Failover chains and health-aware routingIncludedIncludedIncludedIncludedIncluded
Spend ceilings per keyIncludedIncludedIncludedIncludedIncluded
PII redaction and prompt-injection detection (built-in)IncludedIncludedIncludedIncludedIncluded
Guardrails: custom patterns, block mode, per-key policies and budgetsNot includedIncludedIncludedIncludedIncluded
Usage alerts on Slack, webhook or emailNot includedIncludedIncludedIncludedIncluded
Ledger export (CSV)Not includedIncludedIncludedIncludedIncluded
Presets and classifiersNot includedIncludedIncludedIncludedIncluded
Members: invite your team, any number of seatsNot includedNot includedIncludedIncludedIncluded
Client workspaces, reports and membersNot includedNot includedIncludedIncludedIncluded
Per-workspace guardrail policies and budgetsNot includedNot includedIncludedIncludedIncluded
Single sign-on (OIDC or SAML) and SCIM directory syncNot includedNot includedNot includedIncludedIncluded
Priority email supportNot includedNot includedNot includedIncludedIncluded
Private routing tierNot includedNot includedNot includedNot includedIncluded
7-year ledger retentionNot includedNot includedNot includedNot includedIncluded
SLA by agreementNot includedNot includedNot includedNot includedIncluded

What a month costs: a worked example

A team sends $40,000 a month of inference at list price. That $40,000 goes to its providers whichever gateway it uses; what differs is what each one adds on top.

GatewayAdded on top
Autonomous Relay, Team plan (0% markup)$99 a month, $1,188 a year
OpenRouter with your own keys (5% above $25,000 a month)$750 a month
OpenRouter credits, Standard (5.5%)$2,200 a month

OpenRouter’s rates as published on 2026-11-01. Try another amount in the OpenRouter comparison

Questions

Is there a free trial?
Yes. Every new organisation gets 14 days of the Team plan, and no card is needed to start.
What happens when the trial ends?
Inference on your own provider keys, your API keys, Activity, Logs and Members keep working. Features you set up during the trial keep running and stay readable; creating or changing one needs a plan that includes it.
Is a plan priced per seat?
No. Every plan is per organisation. Pro is for one person; Team and Business take any number of members.
Do you charge per token or per request?
No. Markup on inference is 0% and there is no per-request fee; your provider bills you for tokens at its own price.
Is there a contract?
Pro, Team and Business are monthly. Enterprise is an annual, invoiced contract, and that contract governs its terms.

Comparing gateways? See Autonomous Relay vs OpenRouter.