How AISIX Fits

AISIX AI Gateway is a dedicated gateway for AI traffic. It sits between applications and upstream model providers, so application teams can call stable model aliases while platform teams manage provider credentials, routing, policy, and usage controls centrally.

Applications call AISIX with a caller API key and a model alias. AISIX resolves that request to the configured provider credential, upstream model, and traffic policy.

Why Use AISIX

AI traffic often starts with direct provider integration: an application stores a provider key, chooses a provider model ID, and calls that provider's API.

This approach can be sufficient for a single application. It becomes harder to operate when many applications need shared credentials, shared policy, provider failover, observability, or model changes that should not require application redeploys.

AISIX keeps those provider-side decisions behind the gateway:

Use AISIX when AI traffic needs centrally managed credentials, aliases, policy, and provider routing. If one application calls one provider directly and does not need shared keys, shared policy, or provider abstraction, a direct provider integration may be enough.

Operating Model

At request time, AISIX authenticates the caller API key, checks whether that key can use the requested model alias, resolves the alias to provider-side configuration, applies gateway policy, and forwards the provider request upstream.

The main operating pattern is separation of concerns: applications use stable model aliases and caller API keys, while provider choice, provider credentials, and policy stay centralized at the gateway.

AISIX vs AI Plugins

Apache APISIX and API7 Gateway can add AI behavior to existing API gateway routes through AI plugins. That path fits teams whose main workload is API traffic and whose AI calls are a few routes in a broader API gateway deployment.

AISIX is designed for deployments where AI traffic is the primary gateway workload. It treats provider keys, model aliases, caller API keys, routing, rate limits, cache, guardrails, observability, and managed budget checks as gateway resources instead of route-level plugin settings.

Use the APISIX or API7 Gateway plugin path when AI behavior belongs inside existing API gateway routes. Use AISIX when the operating focus is AI traffic itself, including centralized provider access, model access, routing, policy, managed budget checks, and AI request telemetry.