Kong AI Gateway 2.0 is here!
We’ve decoupled AI management from standard API proxying to give you a dedicated runtime built specifically for LLMs, Model Context Protocol (MCP), and agentic workflows.
Here is what’s dropping under the hood:
-
Independent AI Control Plane: Faster release cycles mean instant support for new LLMs and agentic protocols on day one, unlinked from standard API Gateway release trains.
-
First-Class AI Primitives: Native configuration objects for
AI Model,AI MCP Server,AI Agent, andAI Policies;no more translating LLM intent into standard REST routes. -
Semantic Guardrails: Real-time prompt injection protection, jailbreak detection, and content safety controls.
-
Token Budgeting & Cost Control: Enforce token-aware rate limits across teams and export per-request OTLP AI telemetry straight to your observability stack.
-
GitOps-First Migration: Convert existing decK configurations into native 2.0 resources using the updated
kongctlCLI.
Try the local Docker quickstart or spin up the new AI control plane in Konnect: Kong AI Gateway | Kong Docs
Dive deeper on AI Gateway 2.0 concepts - AI Gateway 2.x concepts - AI Gateway | Kong Docs
See a demo! Kong AI Gateway 2.0 Overview | Loom
