Tired of juggling multiple LLM API formats? I built a small gateway to fix it

Working with different model providers is exhausting.
Every vendor has its own request & response schema, error codes, auth setup… I kept rewriting the same adapter code over and over.

So I built a simple gateway to normalize all LLM API calls.
One unified interface, switch between different models without rewriting your code.

This is just a side project I made for my own workflow.
Curious: how do you handle multiple LLM APIs in your projects? Do you roll custom adapters or use existing tools?