LLM Gateway

Point your apps at one OpenAI-compatible endpoint. Every request gets guardrails, routing, limits and cost tracking, whichever provider serves it.

Quick startOpen in console

Console › Settings › AI Gateway › LLM Gateway

Get going

Send your first requestCreate an app, copy its key, call the gatewayConnect a providerOpenAI, Anthropic, Azure, Bedrock, Vertex, OCITurn on guardrailsBlock or redact sensitive data and attacksAutomate apps and keysManage apps, keys and configuration with the Management API

Everything in LLM Gateway

37 pages

Get started5

• Introduction
• Quick start
• Architecture
• Integration guide
• HA and SLA

Apps and providers6

• Applications and keys
• Providers and models
• Provider support matrix
• AWS Bedrock
• Oracle OCI
• SDK mode

Protect7

• Security guardrails
• Custom detections
• Guardian agent
• Hallucination protection
• Identity and network trust
• Gateway access and allowed models
• Tool controls

Cost and traffic4

• Routing and fallbacks
• Rate, token and budget limits
• Token saving
• Prompt store

Monitor4

• Agent monitoring
• Conversation grouping
• Alerts
• App audit log

Self-service3

• Self-service overview
• Admin guide
• Developer guide

API reference7

• Unified completions
• Log export API
• Management API
• Apps and credentials API
• Providers and configuration API
• Prompts API
• Policy API (QuilrQL)

Release notes1

• UX changelog