What does an API management platform actually do?
It sits between the people or systems that call an API and the services that answer. A gateway enforces authentication, rate limits and routing on every request. A developer portal lets consumers find the API, read its documentation and obtain a key. Analytics show who uses what.
Some products bundle all three; others do one well. KrakenD is a gateway first and calls itself stateless, with no database to operate. Apiable does the portal and the commercial layer and deliberately leaves the gateway to another product. Axway and SAP sell the full lifecycle as a suite.
The practical consequence is that "API management" on a vendor page can mean three different purchases. Before comparing prices, write down which of the three you are actually missing.
Why does jurisdiction matter more for an API gateway than for most software?
Because the gateway sees the traffic. Every token, header and payload that crosses the boundary of your organisation passes through it, and a hosted control plane also holds the configuration of every API you expose.
The six vendors here are all established in the EU, and the more important distinction is where the data plane runs. KrakenD states that you keep everything within your own infrastructure. Axway lets gateways run on-premises, in any cloud or both under a central control plane. Apiable, which is a portal rather than a gateway, runs on AWS in Frankfurt and says API traffic never passes through it.
Axway deserves a precise sentence: it is a French company under a French listed parent, 74Software, but it also operates a large office in Scottsdale, Arizona. An organisation with a strict reading of ownership should read that nuance before shortlisting it.
Open-source gateway or enterprise suite?
An open-source gateway is the right default when the team is engineering-led and the requirement is routing, authentication and rate limiting in front of microservices. KrakenD Community Edition is free under Apache 2.0, runs without a database and scales out by adding instances. Fusio is an Apache 2.0 platform that goes the other way, turning databases and business logic into API products with a built-in developer portal.
An enterprise suite earns its price when governance, a formal developer programme and analyst-recognised support matter. Axway cites 10 consecutive years as a Leader in the Gartner Magic Quadrant for API Management. SAP API Management is the pragmatic pick when the estate already runs on SAP S/4HANA and the Business Technology Platform.
The cost of the open-source route is that you assemble the portal, billing and analytics yourself, or buy them from a second vendor.
Is a managed Otoroshi, a self-run gateway or a suite cheaper?
It depends on what you count. A self-run open-source gateway has no licence fee and a real operations cost. Cloud APIM puts a price on that operations work: Otoroshi Managed starts at €250 per month for a dedicated cluster, and the serverless product has a free tier before plans from €30 per month, all excluding VAT.
Suites quote on request. Axway and SAP both state no public price, and KrakenD Enterprise says its price is not linked to the number of APIs or to throughput, which makes it predictable but still a negotiated figure.
For a small team the sensible comparison is a €250 managed cluster against the engineer-hours of running the same open-source software yourself. For a large estate it is a negotiated suite licence against the sum of several smaller products.
Do you need a gateway at all for AI traffic?
Increasingly the same control point is being asked to govern calls to language models. KrakenD 3.0 ships an AI Router and semantic cache, and the KrakenD AI Gateway adds LLM routing, prompt guardrails, token budgets and a built-in MCP server.
Cloud APIM sells AI Studio, an AI gateway that puts LLM providers behind one OpenAI-compatible API with keys, usage and budgets per team, from €49 per month after a free tier. Fusio supports the Model Context Protocol natively.
Treat this as a reason to ask your API vendor one question: whether LLM traffic goes through the same policy engine as your REST traffic, or through a separate product with its own jurisdiction.
