How do you create repeatable tests in a non-deterministic system? Every time you send the same prompt you get a different answer.

This is a good overview of how this is done: https://www.anthropic.com/engineering/demystifying-evals-for...

The actual tokens might be non-deterministic, but you could look for proxy measures that are supposed to be invariant. Eg. correctness/performance on benchmarks, "thinking level" on complex problems, etc

That's like the entire field of statistics.