Skip to main content

About

What Is the Radicalbit AI Gateway​

The Radicalbit AI Gateway is a centralized access point to the generative AI models used across your organization. It sits between your AI applications and the underlying models — analyzing traffic, filtering information based on configurable conditions, and recording metrics and events to give you full visibility into every operation.

The result is a single, controlled layer that makes your AI applications observable, governed, and cost-efficient. The Gateway is application-agnostic and fully compatible with the OpenAI standard, which means it integrates with the vast majority of AI applications and frameworks without requiring any code changes.

Mission​

Our mission is to give organizations full control over their AI usage — making it secure, observable, and cost-efficient — without adding friction to the teams building AI applications.

Capabilities​

Governance & Security​

  • Guardrails — rule-based filters (contains, starts with, ends with, regex), PII detection and masking via Microsoft Presidio, and LLM-as-a-Judge for semantic content evaluation
  • Fallback — automatic failover across models when a provider fails or is unavailable
  • Rate Limiting — cap the number of requests per time window at the route level
  • Token Limiting — cap input and output token consumption per time window
  • Budget Limiting — cap spending based on combined token costs per time window
  • API Key Authentication — every route requires a gateway-issued API key; no direct model access

Cost Control​

  • Exact Caching — serve identical requests from memory without calling the model
  • Semantic Caching — match semantically similar requests using embedding similarity
  • Intelligent Routing — route requests dynamically based on keywords, token length, context length, time of day, budget consumption, ML classifiers, or embedding similarity
  • Cost Dashboard — monitor spending by group and API key from the UI

Observability​

  • Prometheus Metrics — request rate, latency, token usage, cache hits, guardrail triggers, fallback activations, and more, exposed on a dedicated metrics endpoint
  • OpenTelemetry Tracing — end-to-end traces of every request, exportable to any OTLP-compatible backend
  • Event Tracking — UI-level visibility into events per route with time filtering
  • UI Monitoring — web interface for routes, groups, keys, cost trends, and feature status

Multi-Provider Support​

  • Native providers — OpenAI, Anthropic, Google Gemini, DeepSeek, Mistral, Azure OpenAI
  • OpenAI-compatible endpoints — Ollama, vLLM, OpenRouter, and any on-premises deployment

Enterprise Capabilities​

Enterprise Feature

The following capabilities are available exclusively in the Enterprise edition. Contact sales for licensing information.

  • Identity Provider Integration — sync users, groups, and roles from Keycloak (or a custom IDP) into the gateway
  • Role-Based Access Control — three built-in roles (Admin, Builder, Auditor) with granular permissions
  • SSO / OIDC Authentication — OpenID Connect-based single sign-on for the admin interface
  • JWT Token Authentication — users can call gateway endpoints directly with IDP-issued tokens instead of gateway API keys
  • Project-Level User Association — assign users to specific projects with role-scoped visibility