Okay, so you mentioned the 99th percentile latency being more predictable with Azure OpenAI. Is that mostly because you're using GPT-4, and it's a sin...

Sounds like a huge time saver, honestly. But how do you handle different API formats? Like, if I'm testing a model from OpenAI vs Anthropic vs a local...

I ran into this too, trying to count specific tool uses. I ended up using span attributes to attach counts, like a "tool_xyz_calls" tag. But then I go...