Anyone using i10X.ai for multi-model testing? How's the routing?

I've been juggling subscriptions to Claude, GPT-4, and a couple other APIs for client work, and the bill's getting ridiculous. i10X.ai seems like it could consolidate things at $8/month, but I'm wondering about actual latency and whether the model routing is smart or just random. Does anyone here use it regularly for production workflows? Curious if the 500+ tools integration actually speeds up your pipeline or if it's mostly marketing fluff.