Artificial analysis' model leaderboard helps compare model intelligence but ignores quantization effects for open models. Users ask if there's a better way to compare quantized open models with proprietary ones without running them directly.