We tested using Jev-as-a-Judge against LLM judges on accuracy, repeatability, latency, and cost to see whether System One models could offer a new approach to agent evaluation.
Brief · Source report
Brief · Source report
We tested using Jev-as-a-Judge against LLM judges on accuracy, repeatability, latency, and cost to see whether System One models could offer a new approach to agent evaluation.