The premise: in 10 minutes compare a bunch of model/parameter/prompt combos to find the best point on the quality-latency-cost tradeoff.
Enter your data manually or sync from your tracing platform.
~300 models to select from.
Here is a 1 minute Loom: https://www.loom.com/share/39b7cd24166c4e1fafd5e7fbd12a9d4d?...