How to Compare AI Gateway Pricing: Token Rates, BYOK and Total Cost
Separate token rates, subscriptions, and features to find the true cost of each gateway option.
Concentrate.ai
On this page
Compare AI gateway prices using the same workload and the full bill. A token rate alone does not tell you what you will pay.
The useful comparison includes model charges, gateway fees, paid features and the work required to run the service. “No token markup” describes one component. It does not establish that credit purchases, BYOK, subscriptions or support are also free.
Separate the costs
input, output, cached input and any other billable units for the exact model and service tier.
LLM Gateway
Instant access to every model
Change models, providers, add fallbacks, and more by running all AI usage through one platform.
BYOK terms: whether supplying your own provider key changes the gateway charge, including free allowances and overage rules.
Paid features: logging retention, team controls, support or deployment options that require another plan.
Operating costs: compute, storage, upgrades and incident response for self-hosted software.
Keep contracted discounts and promotional credits separate from public list prices. A temporary credit changes the amount due this month; it does not establish a lower ongoing rate.
A worked example
Suppose a fixed workload costs $1,000 at the underlying provider. A hypothetical gateway charging a 5% usage surcharge plus a $100 monthly subscription would cost:
$1,000 + ($1,000 × 5%) + $100 = $1,150.
For $10,000 of the same provider usage, the total would be $10,600. The fixed subscription becomes a smaller share of the bill as usage rises. These are illustrative rates, not quotes for Concentrate or another vendor.
Do not reuse this formula for a fee charged on credit purchases without checking how credits convert into usable balance. Request fees also require a request count; they cannot be calculated from token spend alone.
Make the vendor comparison fair
For Concentrate, OpenRouter, Vercel, LiteLLM, Bifrost and Portkey, record the following against current first-party pricing or a written quote:
The exact model, provider, service tier and input/output mix.
Whether traffic uses provider keys or gateway-managed billing.
Every applicable fee, allowance and minimum commitment.
The features included in the quoted plan.
The date checked and a link to the source.
For self-hosted options, show software charges separately from infrastructure and engineering work. For managed options, distinguish included service from separately priced support. Do not assign an invented hourly maintenance cost to make one option look cheaper.
Include direct provider access
Direct access belongs in the comparison when the application uses one provider and does not need shared routing or access controls. A gateway earns its place when its features or operating model justify the total cost for your team.
Use our gateway comparison to narrow the shortlist, then calculate costs for your own request mix.
Build your comparison with Concentrate
Use Concentrate’s spend tracking to group trial usage by team, project, key, model and provider. Match those requests to the current Concentrate pricing and the terms for your account. That gives you a request-level basis for comparing a managed gateway with direct access or self-hosted software.
Concentrate’s standard pay-as-you-go pricing lists no token markup, service fee or credit-card fee. You can purchase inference through one account or use supported provider credentials. Include any specialized healthcare, support or enterprise terms when comparing the full account cost.
Blog