One API to Route Them All.
Zero Downtime. Multi-Model Power.
OmniRouteAI unifies 50+ Cloud AI Providers, 8 Web Search Engines, and 16+ Local CLI Tools into a single OpenAI-compatible API endpoint with intelligent priority-tier routing, atomic key rotation, circuit breakers, and real-time streaming.
Quick Start & Integration
Drop-in replacement for OpenAI SDK, Python, cURL, or auto-routing
curl -X POST http://localhost:3000/v1/chat/completions \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "omniauto",
"messages": [{"role": "user", "content": "Explain quantum computing in 2 sentences."}]
}'Core Features & Architecture
Production-grade resilience, performance optimization, and universal compatibility built-in.
Intelligent Auto-Routing
Auto-route requests across all active providers using model weight sentinels (omnilowest to omnihighest), priority tiers (0–4), and automatic vision/audio/video capability filtering.
Atomic Lua Key Rotation
Race-free least-used key selection with Redis Lua scripts. Monitors per-key Requests Per Minute (RPM) and automatically disables failing keys after consecutive errors.
Circuit Breakers & Retries
Built-in 3-attempt strict retry logic (Key 1 → Key 2 → Provider failover). Automatic circuit breakers trip failing providers (≥50% error rate) for 5 minutes to ensure high availability.
Multimodal & Audio Engine
Unified /v1/audio/transcriptions endpoint routing across Whisper, Groq, AssemblyAI, and Deepgram. Native vision and video prompt analysis across top models.
8 Integrated Search Engines
Real-time web search grounding powered by Tavily, Brave, Serper, Exa, Firecrawl, SearchAPI, Google PSE, and unlimited free DuckDuckGo search with no API key needed!
SSE Streaming & Deep Reasoning
Full Server-Sent Events (SSE) support with real-time thinking/reasoning token visualization for Claude 3.7, Gemini 2.0, DeepSeek R1, and Grok 3.
Local CLI & Ollama Bridge
Connect 16+ local CLI tools (Claude Code, Gemini CLI, Ollama, Antigravity, Kilo AI, Copilot CLI) into a single proxy daemon with localized authentication.
Semantic Caching & Job Queue
Redis-powered semantic response caching for sub-millisecond responses on repeated queries, combined with BullMQ worker queues for asynchronous non-streaming batches.
Request Processing Lifecycle
How OmniRouteAI handles every incoming request from client to provider response.
Client Request
Standard OpenAI API payload sent to /v1/chat/completions or /v1/audio/transcriptions.
Capability & Router
Extracts model/weight sentinel, inspects multimodal attachments, and samples active priority tier.
Atomic Key Manager
Lua script picks least-used API key, enforces RPM limits, and checks provider circuit breaker status.
Adapter Execution
Transforms request into target provider format (OpenAI, Anthropic, Gemini, Groq, Ollama, CLI bridge).
Streaming & Cache
Delivers SSE chunks with reasoning tokens to client and stores prompt response in Redis semantic cache.
Ecosystem & Provider Network
Seamless connectivity with cloud AI leaders, open-source models, search providers, and local tools.
Dashboard Overview
Provider Health
| Provider | Status | Error Rate | Models | Priority |
|---|---|---|---|---|
| Loading... | ||||
Providers
Model Weights
Set a weight (1–5) for each model. Weight 1 = basic/cheap, weight 5 = premium/best.
Default weight is 3. Use omnilowest–omnihighest models
to auto-route by weight tier.
| # | Provider | Model | Weight | Status |
|---|---|---|---|---|
| Loading models... | ||||
API Keys
| Key | Usage Today | Tokens Today | RPM | Status | Actions |
|---|---|---|---|---|---|
| Select a provider | |||||
Request Logs
| Time | Request ID | Provider | Model | Status | Latency | Tokens | Error Message |
|---|---|---|---|---|---|---|---|
| Loading... | |||||||
Usage Statistics
Token History Graph
Provider Breakdown (Today)
| Provider | Requests | Input Tokens | Output Tokens |
|---|---|---|---|
| Loading... | |||
Daily History
| Date | Total Requests | Input Tokens | Output Tokens | Cost Estimate | |
|---|---|---|---|---|---|
| Loading... | |||||
Search Providers
Add API Key
| Key | Usage | RPM | Status | Actions |
|---|---|---|---|---|
| Select a provider | ||||
AI Playground
🦙 Ollama Playground
Connection Status
Daemon Logs (Ollama)
Loading...
Settings
Backend Connection
Local Daemon Connection
Requests are load-balanced across all daemons. If one is offline, failover is automatic.
Quick Actions
Connection Status
Not tested yet
Local CLI Auth Management
Harvested Sessions
The daemon automatically monitors your local config files (Claude Code, Cursor, GitHub gh, etc.) to borrow active sessions.
| Provider | Method | Status | Expiries In | Actions |
|---|---|---|---|---|
| Loading status... | ||||
Direct OAuth / MITM Settings
MITM_PROXY=true in daemon env allows capturing tokens from keychain-managed tools like Grok.