A feature matrix does not survive a finance review. The internal case for an LLM gateway is four measurable outcomes — total spend, unit cost, performance under load, and shipping speed — each with a number attached.
You could run your own LLM gateway, and some teams should. We chose to run one for you instead. Here is the operational bill self-hosting hands you, the mechanism we replace it with, what our choice genuinely costs you, and the teams that should still self-host.
An LLM gateway is one endpoint in front of every model provider that owns six cross-cutting jobs — auth, cost, limits, safety, observability and failover. What it does, what happens to a request inside it, and when you need one.
TrueFoundry's gateway is one module of a platform that also sells model deployment, GPU serving, and agent, MCP and skills registries — metered by requests and by seat. nRouter sells the gateway alone, priced as a share of model spend. A procedure for deciding which purchase you are making.
Kong's AI plugins put LLM governance in the data plane you already run — and leave you running it: provider credentials in plugin config, Redis behind the rate limiter, SSO and audit logging on the Enterprise tier. nRouter is the other trade: a managed endpoint that holds the provider keys.
Forwarding a request to a provider is a solved problem. Tenant isolation, credits that cannot go negative, guardrails on every call and one bill are not. That asymmetry decided what nRouter is — and what we refuse to ship.
