Skip to main content

Compare · AI gateways and routers

Limitr vs Portkey.

Portkey governs AI traffic at a gateway, now as part of Palo Alto Networks. Limitr decides in your app what each customer may use, and bills them for it.

Start freeBook a demo
The short answer
Portkey

AI gateway, now Palo Alto Networks' Prisma AIRS AI Gateway.

Limitr

Decides every call in your app, with prices, costs, limits, and contract terms, then bills it.

Side by side

What happens when…

The moments that decide it, answered from each product's own docs and pricing.

When…PortkeyLimitr
An agent loop runs away at 2 a.m.Cost and token budgets on provider integrations (Enterprise).The call that would break the budget never runs. Task, customer, and org budgets are checked in your app first.
A customer costs more than they payCost tracking at provider prices.Margin per account is live, and a cap on what an account can cost you stops it before it turns unprofitable.
A customer's AI budget runs lowRouting, fallbacks, and caching follow providers, not the customer’s plan.Your app asks Limitr which models the customer may use and can afford, and switches or downgrades. No proxy, no token resale.
Billing your customers for AINot a billing tool.Meters each customer, keeps the ledger, and sends invoices. Collect through Stripe, or bill manually.
The service is unreachableEvery model call goes through the gateway.Nothing to reach. Decisions run in your process and keep working offline.

Checked November 10, 2026. Sources: Palo Alto Networks completes Portkey acquisition · Portkey budget limits. Something out of date? Tell us.

Architecture

Where the decision happens.

The difference most comparisons skip: where a call is allowed, priced, and capped, and when.

Portkey
A gateway in the request path

Every call routes through their gateway or proxy, which enforces budgets per key at provider prices.

Your app
every call
Gatewaycaps per key
then
Model or APIthe cost
Limitr
Inside your app, before the call

The engine runs natively in your process and decides every call in about a millisecond: allowed, priced at your cost and your price, and held to the contract. Usage goes to Limitr Cloud in the background.

Your appLimitr engine inside
decided first
Model or APIthe cost
in the background
Limitr Cloudledger and invoices

Proof

Don't take our word for it.

Bring last month's usage. We'll replay it through Limitr Analyst and show your margin per account, and what Limitr would have stopped.

100,000+users metered for The Regis Company's Fortune 500 contract971 of 1,000attack calls stopped inside the app in our abuse demo, on a real policy0network calls to decide a request. It keeps deciding when the network doesn't.Apache 2.0Read exactly what decides your calls.
Book a replayStart free

The gap

What you'd still build on Portkey.

Teams pair Portkey with code of their own. With Limitr, that part is the product.

With Portkey, you still build
With Limitr, you get
Limits per customer and plan, not per provider key
Limits tied to the customer's plan and contract
Revenue next to spend, for margin
Your cost and your price on every call, so margin per account is live
Metering and invoices for your customers
Metering, ledger, and invoices for your customers

Choose Portkey for AI traffic security and governance inside Prisma AIRS.

Switching

From Portkey to Limitr.

No migration day. Limitr runs next to what you have until the numbers match.

Book a replayMap your pricing
Keep Portkey for AI traffic security and routing.
Run Limitr in observe mode next to Portkey and compare the numbers on real traffic.
Switch enforcement on. Customers come over as they appear, with no migration day.

Common questions

Limitr vs Portkey, answered.

Yes. Palo Alto Networks completed its acquisition of Portkey in May 2026, and the gateway is now the Prisma AIRS AI Gateway.

Portkey governs AI traffic at a gateway, now as part of Palo Alto Networks. Limitr decides in your app what each customer may use, and bills them for it. Choose Portkey for AI traffic security and governance inside Prisma AIRS.

Portkey: a gateway in the request path. Every call routes through their gateway or proxy, which enforces budgets per key at provider prices. Limitr: inside your app, before the call. The engine runs natively in your process and decides every call in about a millisecond: allowed, priced at your cost and your price, and held to the contract. Usage goes to Limitr Cloud in the background.

Keep Portkey for AI traffic security and routing. Run Limitr in observe mode next to Portkey and compare the numbers on real traffic. Switch enforcement on. Customers come over as they appear, with no migration day.

Also compared

More in aI gateways and routers.

Every comparison →
Limitr vs LiteLLMOpen-source LLM proxy with virtual keys, budgets, and routing.Compare →Limitr vs OpenRouterOne API and one credit balance for hundreds of models, joining Stripe.Compare →

See it on your own pricing.

npm i @formata/limitr
Start freeBook a demo