Multi-key routing with protocol translation and resilient failover.

Showcase a developer-first gateway that speaks OpenAI, Gemini, and Anthropic downstream protocols while balancing traffic across upstream OpenAI-compatible keys with cooldown control, rate-limit handling, and operational visibility.

Launch Admin Console Explore Capabilities

OpenAI Compatible Gemini Adapter Anthropic Adapter

Protocol Modes

OpenAI, Gemini, Anthropic downstream support

Routing Focus

Smart

Weighted selection with cooldown-aware retries

Deployment

Edge

Designed for Cloudflare Workers runtime

UX Style

Tech

Animated but performance-conscious presentation

Core capabilities

The demonstration layer presents the project in a cleaner, more technical style while keeping interactions lightweight and fast. Motion is subtle, reduced-motion safe, and focused on emphasis rather than distraction.

Adaptive key orchestrationDistributes upstream traffic through weighted balancing, concurrency limits, and cooldown recovery logic.

Protocol translationAccepts multiple downstream formats and normalizes them to a single upstream OpenAI-compatible execution path.

Operational resilienceRetries transient failures, tracks request outcomes, and exposes runtime state through the admin console.

Main demonstration endpoints

These are the main public-facing routes exposed by the gateway. The administrator interface remains unchanged and accessible separately.

OpenAI style

POST /v1/chat/completions
POST /v1/responses
GET  /v1/models

Gemini style

GET  /v1beta/models
POST /v1beta/models/{model}:generateContent
POST /v1beta/models/{model}:streamGenerateContent

Anthropic style

POST /v1/messages
GET  /anthropic/v1/models