Overview
Sylica provides a single OpenAI-compatible gateway that routes to multiple model providers behind one API contract.
The primary design goal is to eliminate provider lock-in at application level while still preserving deterministic control when you need it. Your integration sends one request format, and Sylica chooses a suitable provider based on availability, route policy, and requested model strategy.
In practical terms, this means your product team can ship features once, then iterate provider selection independently. You can optimize for speed in interactive flows, optimize for cost in batch workloads, and maintain consistent response parsing logic in both cases.
Integration Model
Most teams integrate using existing OpenAI SDKs and only change two settings: baseURL and API key. The request and response envelopes remain aligned with OpenAI chat completions, so your client-side and server-side parsing code stays simple.
How To Navigate These Docs
Documentation is split by implementation stage so teams can onboard faster and avoid missing critical production details. Read sections in order for new integrations, or jump directly to Routing, Errors, and Production Readiness when hardening existing workloads.