Unified
Intelligence
The Global
Infrastructure
OpenAI Stack
Full support for GPT-5.5, GPT-5.4, GPT-4.1, and legacy endpoints via unified aurora routing.
Anthropic
Native Claude Fable 5 and Opus 4.8 support with optimized header management and routing.
Meta Llama 4
Deploy local Llama 4 weights behind Aurora for private, high-speed inference.
Mistral AI
Optimized Mistral Medium 3.5 and Large 3 routing with region-aware load balancing.
Gemini (Vertex AI)
Vertex AI integration for Gemini 2.5 Pro multimodal tasks and high-context window availability.
Groq
Ultra-low latency inference for Llama 4 Scout, DeepSeek R1, Qwen 2.5, and Whisper via Groq LPU.
DeepSeek
DeepSeek-V3 and DeepSeek-R1 reasoning models routed through Aurora's unified gateway.
OpenRouter
Unified access to 300+ models from every major and emerging provider via one endpoint.
xAI (Grok)
Grok-3 and Grok-3-mini with reasoning and real-time knowledge via xAI API.
Ollama
Run any open-weight model locally — Llama, Qwen, DeepSeek — via Ollama's OpenAI-compatible endpoint.
Azure OpenAI
Deployments of GPT-4.1, GPT-4o, o3, o4-mini on Azure with managed compliance and region pinning.
vLLM
Serve any HuggingFace model with blazing-fast inference via vLLM backend.
Capability
Matrix
| Provider | Streaming | Tool Use | Vision | Lat. (p50) | Aurora Rank |
|---|---|---|---|---|---|
| OpenAI (All)[STABLE] | YES | YES | YES | 8.4ms | S-Tier |
| Anthropic[STABLE] | YES | YES | YES | 12.1ms | S-Tier |
| Mistral[STABLE] | YES | YES | NO | 10.5ms | A-Tier |
| Local Llama[EDGE] | YES | BASIC | NO | 1.2ms | Edge |
| Cohere[STABLE] | YES | YES | NO | 14.2ms | B-Tier |
Custom
Ecosystem
Aurora is built on a modular Go-Plugin architecture. Integrate your internal proprietary LLMs or legacy systems with minimal overhead using our standardized middleware hooks.
Custom GRPC Hooks
Integrate any binary endpoint via high-speed GRPC.
Auth Transformation
Map proprietary auth headers to standard gateway keys.
STATUS: DEPLOYABLE