Overview
Axora is a unified AI gateway. Send one request to /v1/chat and it routes through your configured providers with automatic fallback. Every request is authenticated, rate-limited, and logged.
Base URL
https://api.aivorylabs.in
Auth
Bearer API key
Rate Limit
20 req/min per key
Architecture
Axora is a self-hosted platform with a chat UI, an admin dashboard, and a developer API. A single FastAPI gateway powers all of it — routing LLM requests, serving the web app, and persisting usage to PostgreSQL with Redis behind the scenes. Your app only talks to the gateway; Axora handles callbacks, caching, and failover.
Product Architecture
Clients
FastAPI Gateway
Persistence
LLM Providers
Request Flow
Every endpoint passes through the same middleware and auth layer. Only the LLM-facing routes (/v1/chat, /v1/embeddings ) add RAG context and the fallback chain; the rest read and write data in PostgreSQL and Redis. /v1/chat is just one of many endpoints on the same gateway.
Fallback Chain
If the primary provider fails, Axora automatically retries with the next provider in the chain. Configure your fallback order in the dashboard (or via the fallback-chain endpoints), and use provider_preference to force the primary provider. The response includes provider_used and model fields so you always know which model served the request. When fallback occurs in the built-in Chat UI, an amber warning banner is shown.
| Priority | Provider | Models |
|---|---|---|
| 1 | Groq | llama-3.3-70b-versatile, llama-3.1-8b-instant |
| 2 | OpenRouter | Gemma 4 26B, Gemma 4 31B, Nemotron, GPT-OSS |
| 3 | Gemini | gemini-2.0-flash |
| 4 | OpenAI | gpt-4o, gpt-4o-mini |
| 5 | Claude | claude-3-5-sonnet, claude-3-5-haiku, claude-3-opus |
Authentication
All requests require a valid API key in the Authorization header using Bearer token format.
Header
Authorization: Bearer ax_live_YOUR_API_KEYKeys use the ax_live_ prefix and are scoped to your account. Generate and manage keys in your API Keys dashboard.