AI Gateway.
Route every AI request through the Arbitex content security engine. 9+ LLM providers, three routing modes, and the full 3-tier DLP pipeline on every prompt and response.
Every AI request passes through the engine.
The AI Gateway module accepts OpenAI-compatible API calls and routes them to the configured provider. Before any request reaches a model, the platform engine inspects the content, evaluates policy, and logs the decision. The same engine that protects email and file uploads protects your AI traffic.
base_url = "https://api.arbitex.ai/v1"That is the entire integration. Every request now flows through the content security engine.
What the AI Gateway does.
9+ LLM Providers
OpenAI, Anthropic, Google Gemini, Mistral, Cohere, AWS Bedrock, Azure OpenAI, Groq, Ollama, plus BYOE for self-hosted models. Bring your own provider credentials — register your own API keys, stored with envelope encryption.
Three Routing Modes
Single model, side-by-side compare, or fan-out summarize. Switch modes per request without changing your integration.
Streaming Support
Full streaming pass-through with real-time DLP inspection. No buffering delays. Responses stream to the client as they arrive from the provider.
Cost Controls
Budget caps, usage quotas, and per-tenant cost analytics. Set dollar limits per team, project, or model with hard enforcement.
Model Health Monitoring
Circuit breakers, automatic failover, and latency tracking. Kill switch for instant model lockout across all routing.
Full DLP on Every Request
The platform engine inspects every prompt and response through all 3 DLP tiers. Compliance packs enforce framework-specific rules. Actions are deterministic.
Govern every AI call.
See how the AI Gateway channel connects to the Arbitex content security engine.