01 / ROUTE
Stable model aliases
Point applications at names such as chat-default and change the backing model without shipping new client configuration.
Production model routing
One endpoint.
Every model.
Route OpenAI-compatible requests across model providers with one API, predictable budgets, and automatic failover.
OpenAI SDK compatible · Streaming supported · Per-key controls
quickstart.py
from openai import OpenAI client = OpenAI( base_url="https://gateway.example/v1", api_key="sk-..." ) response = client.chat.completions.create( model="chat-default", messages=[{ "role": "user", "content": "Summarize this incident." }], max_tokens=300 ) print(response.choices[0].message.content)
OpenAI compatibleDrop-in SDK base URL
StreamingServer-sent events
Key budgetsUsage and rate controls
Automatic routingAliases and fallbacks
A smaller gateway surface
Routing without rebuilding your application.
Keep one client integration while model aliases, provider selection, and access policy evolve behind the endpoint.
02 / CONTROL
Keys with boundaries
Issue project credentials with model access, request limits, and budgets defined at the gateway.
03 / OBSERVE
Request-level visibility
Review latency, status, token use, and routing outcomes from one consistent operational surface.
Model catalog
Use the names your code already knows.
Provider-specific identifiers and fallbacks stay behind a compact OpenAI-compatible API.
Gateway operational
chat-defaultGeneral chat and extraction
availablegpt-6-astraFrontier reasoning alias
availablegpt-6-solGeneral reasoning alias
availablegpt-6-lunaFast inference alias
availableclaude-opus-5-5Compatibility alias
availableclaude-sonnet-5-5Compatibility alias
availablegpt-4o-miniCompatibility alias
availableclaude-3-5-sonnetCompatibility alias
availablegemini-2.0-flashCompatibility alias
available